[ { "id": 1456, "name": "Shipamax", "slug": "shipamax", "former_names": [], "small_logo_thumb_url": "https://bookface-images.s3.amazonaws.com/small_logos/0485262b3034147c1f753ae66224e32cd01bffa2.png", "website": "https://shipamax.com", "all_locations": "London, England, United Kingdom", "long_description": "Shipamax's core technology can 'understand' any logistics document - outputting a business-ready, machine-readable format. \r\n\r\nWhy does this matter? When customers order online - it feels smooth & simple. But in the background it's chaos. When goods are physically moved around the world, there can be up to 20 companies involved in the transaction. Currently, this is a complex, messy process, with different systems and teams creating a mountain of manual work to get the job done.\r\n\r\nCustomers value our technology as we to help make these behind-the-sceens processes simple.\r\n\r\nWe're making one of the largest segments of the global economy – logistics – more scalable and adaptable to today's on-demand world.", "one_liner": "Simple, end-to-end automation for logistics orgs", "team_size": 32, "industry": "B2B", "subindustry": "B2B -> Supply Chain and Logistics", "launched_at": 1478141416, "tags": [ "Robotic Process Automation", "Logistics", "NLP" ], "tags_highlighted": [], "top_company": false, "isHiring": false, "nonprofit": false, "batch": "Winter 2017", "status": "Acquired", "industries": [ "B2B", "Supply Chain and Logistics" ], "regions": [ "United Kingdom", "Europe", "Remote", "Partly Remote" ], "stage": "Growth", "app_video_public": false, "demo_day_video_public": false, "app_answers": null, "question_answers": null, "url": "https://www.ycombinator.com/companies/shipamax", "api": "https://devasheeshg.github.io/yc-api/batches/winter-2017/shipamax.json", "year_founded": 2016, "city": "London", "country": "GB", "linkedin_url": "https://www.linkedin.com/company/shipamax/", "x_url": "https://twitter.com/shipamax", "fb_url": null, "cb_url": "https://www.crunchbase.com/organization/shipamax", "github_url": "https://github.com/shipamax", "logo_url": "https://bookface-images.s3.amazonaws.com/small_logos/0485262b3034147c1f753ae66224e32cd01bffa2.png", "app_video_url": null, "dday_video_url": null, "company_photos": [], "primary_partner": null, "founders": [ { "user_id": 126766, "full_name": "Jenna Brown", "title": "Founder/CEO", "founder_bio": "", "is_active": true, "linkedin_url": "https://uk.linkedin.com/in/jenna-brown-2539b716, https://www.linkedin.com/in/fabian-blaicher-86936120/", "x_url": "", "avatar_thumb_url": "https://bookface-images.s3.us-west-2.amazonaws.com/avatars/9f4fd4628fe946e0fd57fef8071f3fb321ce7586.jpg" } ], "jobs": [], "news": [ { "title": "WiseTech Global buys data entry automation business Shipamax", "url": "https://www.stattimes.com/logistics/wisetech-global-buys-data-entry-automation-business-shipamax-1346913", "date": "Nov 02, 2022" } ], "launches": [] }, { "id": 1873, "name": "Warpdrive by Synthetic Minds", "slug": "synthetic-minds", "former_names": [ "XXV Inc.", "Synthetic Minds" ], "small_logo_thumb_url": "https://bookface-images.s3.amazonaws.com/small_logos/a8bc972f9200404d5b7bdf30cf4fd37ff60f4653.png", "website": "http://warpdrive.co", "all_locations": "San Francisco, CA, USA; Remote", "long_description": "What if you could turn your screen recording into an unattended automation, or on-screen guides for your coworkers? That is exactly what we allow you to do. We take the recording, run a compiler to analyze it, synthesize a program from that analysis, and create an automation or guide within seconds.", "one_liner": "From screen recordings to desktop automations using program synthesis", "team_size": 7, "industry": "B2B", "subindustry": "B2B -> Productivity", "launched_at": 1513364572, "tags": [ "Hard Tech", "Robotic Process Automation" ], "tags_highlighted": [], "top_company": false, "isHiring": false, "nonprofit": false, "batch": "Summer 2018", "status": "Inactive", "industries": [ "B2B", "Productivity" ], "regions": [ "United States of America", "America / Canada", "Remote", "Fully Remote" ], "stage": "Early", "app_video_public": false, "demo_day_video_public": false, "app_answers": null, "question_answers": null, "url": "https://www.ycombinator.com/companies/synthetic-minds", "api": "https://devasheeshg.github.io/yc-api/batches/summer-2018/synthetic-minds.json", "year_founded": 2018, "city": "San Francisco", "country": "US", "linkedin_url": "", "x_url": "https://twitter.com/_syntheticminds", "fb_url": null, "cb_url": "https://www.crunchbase.com/organization/synthetic-minds", "github_url": null, "logo_url": "https://bookface-images.s3.amazonaws.com/small_logos/a8bc972f9200404d5b7bdf30cf4fd37ff60f4653.png", "app_video_url": null, "dday_video_url": null, "company_photos": [], "primary_partner": null, "founders": [ { "user_id": 64430, "full_name": "Saurabh Srivastava", "title": "Founder", "founder_bio": "Work on building code capabilities for Nemotron at NVIDIA. Primarily focus is on pre-training, evals and data; some post-training.\n\nPreviously:\n* Head of Code Research @ Essential AI (Ashish Vaswani's co)\n* YC W15 and S18: Code synthesis applied to bio and desktop automations respectively.\n* PhD / Postdoc: Code synthesis using theorem provers.", "is_active": true, "linkedin_url": "https://www.linkedin.com/in/saurabh-sriv/", "x_url": "https://x.com/_saurabh", "avatar_thumb_url": "https://bookface-images.s3.us-west-2.amazonaws.com/avatars/372a965319497524c71bec098dca39064071239f.jpg" } ], "jobs": [], "news": [], "launches": [] }, { "id": 11961, "name": "Sorting Robotics", "slug": "sorting-robotics", "former_names": [ "Robotic Sorting Solutions" ], "small_logo_thumb_url": "https://bookface-images.s3.amazonaws.com/small_logos/2dfc5eeb06e8051af24bd5a9c7758dbe219fba21.png", "website": "https://www.sortingrobotics.com/", "all_locations": "Los Angeles, CA, USA", "long_description": "", "one_liner": "We create automation technology for the cannabis industry.", "team_size": 12, "industry": "B2B", "subindustry": "B2B -> Supply Chain and Logistics", "launched_at": 1541124183, "tags": [ "Robotic Process Automation", "Robotics", "Computer Vision" ], "tags_highlighted": [], "top_company": false, "isHiring": true, "nonprofit": false, "batch": "Winter 2019", "status": "Active", "industries": [ "B2B", "Supply Chain and Logistics" ], "regions": [ "United States of America", "America / Canada", "Remote", "Partly Remote" ], "stage": "Early", "app_video_public": false, "demo_day_video_public": false, "app_answers": null, "question_answers": null, "url": "https://www.ycombinator.com/companies/sorting-robotics", "api": "https://devasheeshg.github.io/yc-api/batches/winter-2019/sorting-robotics.json", "year_founded": 2018, "city": "Los Angeles", "country": "US", "linkedin_url": "https://www.linkedin.com/company/sortingrobotics/", "x_url": "", "fb_url": null, "cb_url": null, "github_url": null, "logo_url": "https://bookface-images.s3.amazonaws.com/small_logos/2dfc5eeb06e8051af24bd5a9c7758dbe219fba21.png", "app_video_url": null, "dday_video_url": null, "company_photos": [], "primary_partner": null, "founders": [], "jobs": [ { "id": 88822, "title": "Senior Robotics Software Engineer", "url": "https://www.ycombinator.com/companies/sorting-robotics/jobs/7RRveqW-senior-robotics-software-engineer", "location": "Los Angeles, CA, US", "type": "Full-time", "role": "Engineering", "role_type": "Robotics", "salary_range": "$140K - $180K", "equity_range": "0.15% - 0.30%", "experience": "6+ years", "visa": "US citizen/visa only", "skills": [ "Amazon Web Services (AWS)", "Python", "Linux", "Machine Learning" ] }, { "id": 97722, "title": "Mechatronics Engineer, VLA Gripping & Assembly (Lead)", "url": "https://www.ycombinator.com/companies/sorting-robotics/jobs/GFPB2XH-mechatronics-engineer-vla-gripping-assembly-lead", "location": "Los Angeles, CA, US", "type": "Full-time", "role": "Engineering", "role_type": "Robotics", "salary_range": "$170K - $190K", "equity_range": "0.20% - 1.00%", "experience": "6+ years", "visa": "US citizen/visa only", "skills": [ "C++", "Python", "Machine Learning", "CAD Design" ] } ], "news": [], "launches": [] }, { "id": 12092, "name": "Traverse Technologies", "slug": "traverse-technologies", "former_names": [], "small_logo_thumb_url": "https://bookface-images.s3.amazonaws.com/small_logos/974ac0591edce5b3e32f3208ef9dcf2c5d0a6918.png", "website": "https://traverse.ai", "all_locations": "Singapore, Singapore", "long_description": "Where are the best places in the world to build a wind farm or a hydropower dam? \r\n\r\nTraverse is an AI driven professional engineering firm focused on answering these types of questions.\r\n\r\nWe use AI, high performance computing and automation to speed up traditional prospecting, feasibility assessment and front end engineering work by 10x. Traverse can prospect for exceptionally high ROI project configurations across entire countries and also optimize the returns of individual projects amongst billions of potential permutations, helping our customers eliminate development risks dramatically.", "one_liner": "AI driven professional engineering firm", "team_size": 6, "industry": "B2B", "subindustry": "B2B -> Operations", "launched_at": 1541714572, "tags": [ "Robotic Process Automation", "B2B", "Climate", "Energy", "Renewable Energy" ], "tags_highlighted": [], "top_company": false, "isHiring": false, "nonprofit": false, "batch": "Winter 2019", "status": "Inactive", "industries": [ "B2B", "Operations" ], "regions": [ "Singapore", "Southeast Asia" ], "stage": "Early", "app_video_public": false, "demo_day_video_public": false, "app_answers": null, "question_answers": null, "url": "https://www.ycombinator.com/companies/traverse-technologies", "api": "https://devasheeshg.github.io/yc-api/batches/winter-2019/traverse-technologies.json", "year_founded": null, "city": "Singapore", "country": "SG", "linkedin_url": "", "x_url": "https://twitter.com/traverseAI", "fb_url": null, "cb_url": "https://www.crunchbase.com/organization/traverse-technologies", "github_url": null, "logo_url": "https://bookface-images.s3.amazonaws.com/small_logos/974ac0591edce5b3e32f3208ef9dcf2c5d0a6918.png", "app_video_url": null, "dday_video_url": null, "company_photos": [], "primary_partner": null, "founders": [ { "user_id": 290053, "full_name": "Eugene Huang", "title": "Founder", "founder_bio": null, "is_active": false, "linkedin_url": "https://www.linkedin.com/in/eugenehuangsg/", "x_url": null, "avatar_thumb_url": "https://bookface-images.s3.us-west-2.amazonaws.com/avatars/029835dfa12b4b23d5b28660b96b1f9d5dfbf2dd.jpg" }, { "user_id": 289504, "full_name": "Thet Lin Thu", "title": "Founder", "founder_bio": null, "is_active": false, "linkedin_url": "https://www.linkedin.com/in/thetlinthu/", "x_url": null, "avatar_thumb_url": "https://bookface-images.s3.us-west-2.amazonaws.com/avatars/2d1974b6c339bd6b8265501c7c3cb32c7d22ea67.jpg" } ], "jobs": [], "news": [], "launches": [] }, { "id": 12613, "name": "Dashblock", "slug": "dashblock", "former_names": [ "Datap", "Dashblock (was Datap)" ], "small_logo_thumb_url": "https://bookface-images.s3.amazonaws.com/small_logos/51f8695f18bceaae54b2384727e02f76debf98df.png", "website": "https://dashblock.com", "all_locations": "Paris, Île-de-France, France", "long_description": "Dashblock is the easiest way to access websites programmatically and collect structured data.", "one_liner": "Turn any website into an API", "team_size": 3, "industry": "B2B", "subindustry": "B2B -> Engineering, Product and Design", "launched_at": 1556596578, "tags": [ "Developer Tools", "Machine Learning", "Robotic Process Automation" ], "tags_highlighted": [], "top_company": false, "isHiring": false, "nonprofit": false, "batch": "Summer 2019", "status": "Inactive", "industries": [ "B2B", "Engineering, Product and Design" ], "regions": [ "France", "Europe", "Remote", "Partly Remote" ], "stage": "Early", "app_video_public": false, "demo_day_video_public": false, "app_answers": null, "question_answers": null, "url": "https://www.ycombinator.com/companies/dashblock", "api": "https://devasheeshg.github.io/yc-api/batches/summer-2019/dashblock.json", "year_founded": 2019, "city": "Paris", "country": "FR", "linkedin_url": "https://www.linkedin.com/company/dashblock", "x_url": "https://twitter.com/DashblockHQ", "fb_url": "https://www.facebook.com/DashblockHQ", "cb_url": null, "github_url": "https://github.com/dashblock", "logo_url": "https://bookface-images.s3.amazonaws.com/small_logos/51f8695f18bceaae54b2384727e02f76debf98df.png", "app_video_url": null, "dday_video_url": null, "company_photos": [], "primary_partner": null, "founders": [ { "user_id": 332388, "full_name": "Maxime Corbani", "title": "Founder", "founder_bio": "After studying engineering between France and Brazil, I worked as a consultant in different European countries, founded a consulting firm in Paris (Symolia), and finally started Dashblock to offer an RPA solution to startups.", "is_active": false, "linkedin_url": "https://www.linkedin.com/in/maxime-corbani-10664ba0/", "x_url": "https://twitter.com/mxcrbn", "avatar_thumb_url": "https://bookface-images.s3.us-west-2.amazonaws.com/avatars/08aca2ae6726da3038181872a4a24410f62c0384.jpg" }, { "user_id": 331634, "full_name": "Hugues Pouillot", "title": "Founder", "founder_bio": "Product Engineer", "is_active": false, "linkedin_url": "https://www.linkedin.com/in/huguespouillot/", "x_url": "", "avatar_thumb_url": "https://bookface-images.s3.us-west-2.amazonaws.com/avatars/03ceb8df479770b2696c6c8a5d1a2a8a9bd5ed37.jpg" } ], "jobs": [], "news": [], "launches": [] }, { "id": 13483, "name": "ElectroNeek", "slug": "electroneek", "former_names": [ "ElectroNeek Robotics", "ElectroNeek", "ElectroNeek RPA" ], "small_logo_thumb_url": "https://bookface-images.s3.amazonaws.com/small_logos/60c6f1e1102860aaf28ab55f675c131d4edc4cb3.png", "website": "https://electroneek.com", "all_locations": "Austin, TX, USA; Remote", "long_description": "We help Enterprises transform their business processes with RPA workflows and enable structured data extraction from any document types (Intelligent Document Processing, IDP). Awesome YC deals for founders (check out us in 'Deals'). Connect any system, automate any process. ", "one_liner": "Robotic Process Automation and Document Processing Platform", "team_size": 175, "industry": "B2B", "subindustry": "B2B", "launched_at": 1578613395, "tags": [ "Developer Tools", "Robotic Process Automation", "B2B" ], "tags_highlighted": [], "top_company": false, "isHiring": false, "nonprofit": false, "batch": "Winter 2020", "status": "Active", "industries": [ "B2B" ], "regions": [ "United States of America", "America / Canada", "Remote", "Fully Remote" ], "stage": "Growth", "app_video_public": false, "demo_day_video_public": false, "app_answers": null, "question_answers": null, "url": "https://www.ycombinator.com/companies/electroneek", "api": "https://devasheeshg.github.io/yc-api/batches/winter-2020/electroneek.json", "year_founded": 2019, "city": "Austin", "country": "US", "linkedin_url": "https://www.linkedin.com/company/12989909", "x_url": "https://twitter.com/electroNeek/", "fb_url": "https://www.facebook.com/ELECTRONEEKRPA/", "cb_url": "https://www.crunchbase.com/organization/electroneek", "github_url": null, "logo_url": "https://bookface-images.s3.amazonaws.com/small_logos/60c6f1e1102860aaf28ab55f675c131d4edc4cb3.png", "app_video_url": null, "dday_video_url": null, "company_photos": [], "primary_partner": null, "founders": [ { "user_id": 405697, "full_name": "Sergey Yudovskiy", "title": "Founder/CEO", "founder_bio": "2x YC Founder. Building Adentris to bring AI-driven systems into healthcare operations - a personal mission inspired by the legacy of my fully medical family. As a mathematician, I bring a data-driven, systems approach to product development and GTM strategy.\n\nPreviously, I led ElectroNeek (W20) as CEO, scaling it to ~$5M ARR. Built a global B2B distribution and partnership network across 30+ countries (North & South America, Europe, Asia), and disrupted the RPA market with a novel GTM playbook.", "is_active": true, "linkedin_url": "https://linkedin.com/in/yudovskiy", "x_url": "https://twitter.com/SergeyYudovskiy", "avatar_thumb_url": "https://bookface-images.s3.us-west-2.amazonaws.com/avatars/7da84bfdf132133d48f3c0f5b778cfd2c6a6a515.jpg" } ], "jobs": [], "news": [ { "title": "ElectroNeek is Named One of 20 Rising Stars as Part of Forbes Cloud 100 List | ElectroNeek", "url": "https://electroneek.com/newsroom/awards-and-recognition/electroneek-is-named-one-of-20-rising-stars-as-part-of-forbes-cloud-100-list/", "date": "Dec 18, 2023" }, { "title": "MSP Bet Works Out for ElectroNeek in 2021: The Only Hyperautomation Vendor for IT Service Providers Reports Hypergrowth", "url": "https://finance.yahoo.com/news/msp-bet-works-electroneek-2021-174100627.html", "date": "Feb 02, 2022" }, { "title": "ElectroNeek Raises A $20 Million Series A To Drive Mass RPA Industry Adoption", "url": "https://www.forbes.com/sites/frederickdaso/2021/06/01/electroneek-raises-a-20m-series-a-to-drive-mass-rpa-industry-adoption/?sh=1aa742f217a7", "date": "Jan 06, 2021" } ], "launches": [] }, { "id": 22742, "name": "Axiom.ai", "slug": "axiom-ai", "former_names": [ "Axiom" ], "small_logo_thumb_url": "https://bookface-images.s3.amazonaws.com/small_logos/48ab4ee91faa4c6436804bf3d71a9e06873ff688.png", "website": "https://axiom.ai", "all_locations": "London, England, United Kingdom; Remote", "long_description": "Axiom.ai is building a no-code browser automation tool. This lets anyone automate repetitive work by recording clicking and typing actions in their browser.\r\n\r\nWe think the market has huge potential - companies like UiPath built $10 billion+ businesses by automating processes on the desktop - an industry now known as Robotic Process Automation (RPA). We think RPA has a lot more potential outside the Enterprise. \r\n\r\nAxiom is on its way to being the industry-standard RPA tool for the browser. It now ranks alongside Selenium - the historical industry standard - as the top result when searching for ‘Browser Automation’ on Google, and is the top-result on Chrome’s extension store.\r\n\r\n", "one_liner": "No-code Browser RPA", "team_size": 8, "industry": "B2B", "subindustry": "B2B -> Operations", "launched_at": 1614621114, "tags": [ "Robotic Process Automation", "SaaS" ], "tags_highlighted": [], "top_company": false, "isHiring": false, "nonprofit": false, "batch": "Winter 2021", "status": "Active", "industries": [ "B2B", "Operations" ], "regions": [ "United Kingdom", "Europe", "Remote", "Fully Remote" ], "stage": "Early", "app_video_public": false, "demo_day_video_public": false, "app_answers": null, "question_answers": null, "url": "https://www.ycombinator.com/companies/axiom-ai", "api": "https://devasheeshg.github.io/yc-api/batches/winter-2021/axiom-ai.json", "year_founded": 2019, "city": "London", "country": "GB", "linkedin_url": "https://www.linkedin.com/company/11860018", "x_url": "https://twitter.com/AxiomRPA", "fb_url": "https://www.facebook.com/browserbots", "cb_url": null, "github_url": null, "logo_url": "https://bookface-images.s3.amazonaws.com/small_logos/48ab4ee91faa4c6436804bf3d71a9e06873ff688.png", "app_video_url": null, "dday_video_url": null, "company_photos": [], "primary_partner": { "name": "Jared Friedman", "url": "https://www.ycombinator.com/people/jared-friedman" }, "founders": [ { "user_id": 233316, "full_name": "Yaseer Sheriff", "title": "Founder", "founder_bio": "My co-founders and I ran a software consultancy together, which evolved to focus on RPA. \n\nAxiom is a spin-out of that software consultancy and our previous RPA expertise. \n\nMy background before that was engineering, computer science, automated theorem proving and computational biology (the latter two are surprisingly related). \n\nhttps://yaseersheriff.github.io/", "is_active": true, "linkedin_url": "https://www.linkedin.com/in/yaseer-sheriff-14752b18/", "x_url": "", "avatar_thumb_url": "https://bookface-images.s3.us-west-2.amazonaws.com/avatars/362950285846090d30c22bbfce09e31dad4f5c84.jpg" }, { "user_id": 248498, "full_name": "Simon Williams", "title": "Founder", "founder_bio": "I studied Psychology at university but then switched to programming afterwards when I saw that you could get paid for making things on the computer. This makes me a technical person who is also intersted in people and their many odd ways.\n\nFounded and ran a moderately successul software engineering and web design consultancy for about 8 years. My co-founders and I founded Axiom after transitioning from that business.\n\nI have a big passion for game design theory and making games, too.", "is_active": true, "linkedin_url": "https://uk.linkedin.com/in/simon-williams-20653618", "x_url": "", "avatar_thumb_url": "https://bookface-images.s3.us-west-2.amazonaws.com/avatars/2f111a66f0e84075045624fe82f139557b5984ff.jpg" }, { "user_id": 571653, "full_name": "Alex Barlow", "title": "Founder", "founder_bio": "", "is_active": true, "linkedin_url": "https://www.linkedin.com/in/alex-barlow-9b489330/", "x_url": "https://twitter.com/BarlovianAlex", "avatar_thumb_url": "https://bookface-images.s3.us-west-2.amazonaws.com/avatars/633b9ebd183c1f33e40cb8615b2ade85c5876cd0.jpg" } ], "jobs": [], "news": [], "launches": [] }, { "id": 22958, "name": "Text Blaze", "slug": "text-blaze", "former_names": [], "small_logo_thumb_url": "https://bookface-images.s3.amazonaws.com/small_logos/3155ba3b687741c19a97719f6f226e3e148c98f5.png", "website": "https://blaze.today", "all_locations": "San Francisco, CA, USA; Remote", "long_description": "Text Blaze saves time and increases comms quality by automating repeated tasks. Users save smart templates and insert them anywhere on the web using keyboard shortcuts. Templates can include placeholders, logic, dynamic fields, integrations with other apps, and more. ", "one_liner": "Slash commands anywhere", "team_size": 9, "industry": "B2B", "subindustry": "B2B -> Productivity", "launched_at": 1615472232, "tags": [ "Robotic Process Automation", "SaaS", "B2B" ], "tags_highlighted": [], "top_company": false, "isHiring": true, "nonprofit": false, "batch": "Winter 2021", "status": "Active", "industries": [ "B2B", "Productivity" ], "regions": [ "United States of America", "America / Canada", "Remote", "Fully Remote" ], "stage": "Early", "app_video_public": false, "demo_day_video_public": false, "app_answers": null, "question_answers": null, "url": "https://www.ycombinator.com/companies/text-blaze", "api": "https://devasheeshg.github.io/yc-api/batches/winter-2021/text-blaze.json", "year_founded": 2019, "city": "San Francisco", "country": "US", "linkedin_url": "https://www.linkedin.com/company/text-blaze/", "x_url": "https://twitter.com/text_blaze", "fb_url": "https://www.facebook.com/textblaze", "cb_url": null, "github_url": null, "logo_url": "https://bookface-images.s3.amazonaws.com/small_logos/3155ba3b687741c19a97719f6f226e3e148c98f5.png", "app_video_url": null, "dday_video_url": null, "company_photos": [], "primary_partner": { "name": "Jared Friedman", "url": "https://www.ycombinator.com/people/jared-friedman" }, "founders": [ { "user_id": 116722, "full_name": "Dan Barak", "title": "Founder", "founder_bio": "", "is_active": true, "linkedin_url": "https://www.linkedin.com/in/barakdan/", "x_url": "https://twitter.com/dan_barak", "avatar_thumb_url": "https://bookface-images.s3.us-west-2.amazonaws.com/avatars/f1df1590b7947bbc169a7f69b00a7762ba6700ea.jpg" }, { "user_id": 57708, "full_name": "Scott Fortmann-Roe", "title": "Founder", "founder_bio": "\r\n", "is_active": true, "linkedin_url": "https://www.linkedin.com/in/scottfr/", "x_url": "", "avatar_thumb_url": "https://bookface-images.s3.us-west-2.amazonaws.com/avatars/9b0fcb3eb6deb72c1c19414f6b78bd5afbe1c4c4.jpg" } ], "jobs": [ { "id": 97116, "title": "Product Marketing", "url": "https://www.ycombinator.com/companies/text-blaze/jobs/CidfA3r-product-marketing", "location": "San Francisco / Remote (US)", "type": "Full-time", "role": "Marketing", "role_type": null, "salary_range": "$80K - $150K", "equity_range": "0.50%", "experience": "Any (new grads ok)", "visa": "US citizen/visa only", "skills": [ "Branding", "Growth design", "Marketing design" ] } ], "news": [], "launches": [ { "id": 83737, "title": "AI Blaze - Your personal AI co-pilot", "tagline": "An AI co-pilot that is actually useful for everyone.", "body": "Install AI Blaze [here](https://chromewebstore.google.com/detail/ai-blaze-instantly-use-ai/cebmnlammjhancocbbnfcglifgdpfejc). It’s free!\n\nHere are a few things you can do with AI Blaze (very partial list:)\n\n1. Quickly write an excellent cover letter for a specific role.\n2. Turn your messy notes into a professional-looking and actionable summary.\n3. Summarize the key points in an article.\n4. Write witty comments on social threads.\n5. Translate your email to any language.\n\n![uploaded image](/media/?type=post&id=83737&key=user_uploads/116722/c2c9846e-f5d2-431f-820c-c8e45efed8ea)\n\nAI Blaze allows you to:\n\n1. Create personal AI prompt templates and use them anywhere.\n2. Automatically include the full content of the page in your prompt.\n3. Write new text or improve the written text.\n\n![uploaded image](/media/?type=post&id=83737&key=user_uploads/116722/2b9e8044-b4c6-44a9-a304-44a15a2c6b95)\n\n![uploaded image](/media/?type=post&id=83737&key=user_uploads/116722/a5df25bb-9311-426c-8739-d4338b3a2fcb)\n\n", "url": "https://www.ycombinator.com/launches/Lmb-ai-blaze-your-personal-ai-co-pilot", "votes": 9, "created_at": "2024-09-04T12:11:51.222Z" } ] }, { "id": 23121, "name": "BotBuilt", "slug": "botbuilt", "former_names": [ "ARC:Advanced Robotics Construction", "ARC: Advanced Robotic Construction", "ARC", "ARC: Advanced Robotic Construction" ], "small_logo_thumb_url": "https://bookface-images.s3.amazonaws.com/small_logos/6c0f70c9a27028dd4b634603cfcc5021fc68ccb3.png", "website": "https://www.botbuilt.com", "all_locations": "Southern Pines, NC, USA", "long_description": "BotBuilt is creating flexible robotic systems to solve the housing crisis. Our cutting-edge software and cost-efficient hardware allow us to improve the world by providing beautiful construction, safer job sites, and sustainable building techniques. The $600 billion residential construction industry is facing a massive labor shortage. Our robotic systems leverage the latest in rapid prototyping, artificial intelligence, and computer vision. By taking on some of the hardest technical challenges on earth, BotBuilt is ready to help solve one of society’s biggest problems.\r\n", "one_liner": "We use robots to build houses.", "team_size": 8, "industry": "Industrials", "subindustry": "Industrials -> Manufacturing and Robotics", "launched_at": 1660682183, "tags": [ "Robotic Process Automation", "Robotics", "Construction", "Housing" ], "tags_highlighted": [], "top_company": false, "isHiring": true, "nonprofit": false, "batch": "Winter 2021", "status": "Active", "industries": [ "Industrials", "Manufacturing and Robotics" ], "regions": [ "United States of America", "America / Canada" ], "stage": "Growth", "app_video_public": false, "demo_day_video_public": false, "app_answers": null, "question_answers": null, "url": "https://www.ycombinator.com/companies/botbuilt", "api": "https://devasheeshg.github.io/yc-api/batches/winter-2021/botbuilt.json", "year_founded": 2020, "city": "Southern Pines", "country": "US", "linkedin_url": "https://www.linkedin.com/company/botbuilt", "x_url": "", "fb_url": null, "cb_url": null, "github_url": null, "logo_url": "https://bookface-images.s3.amazonaws.com/small_logos/6c0f70c9a27028dd4b634603cfcc5021fc68ccb3.png", "app_video_url": null, "dday_video_url": null, "company_photos": [], "primary_partner": { "name": "Aaron Epstein", "url": "https://www.ycombinator.com/people/aaron-epstein" }, "founders": [ { "user_id": 550374, "full_name": "Brent Wadas", "title": "Founder", "founder_bio": "From Scottsdale, AZ. Always hungry. Always creating. Grew up in music and business. Started an RIA and a SaaS company while in college, sold out, became an Army officer. Worked in Special Operations, had a blast. Civilian again, grad school at Georgetown for technology and business, consulting for the Sensitive Activities community. Developed BotBuilt with my co-founders. We're here to change the world. ", "is_active": true, "linkedin_url": "https://www.linkedin.com/in/brentwadas/", "x_url": "https://twitter.com/The_Brent_Wadas", "avatar_thumb_url": "https://bookface-images.s3.us-west-2.amazonaws.com/avatars/1410f80f01d87f0993c0423fa7259ea63a4331b3.jpg" } ], "jobs": [ { "id": 63231, "title": "Software Infrastructure Engineer", "url": "https://www.ycombinator.com/companies/botbuilt/jobs/ssohMWR-software-infrastructure-engineer", "location": "Durham, NC, US", "type": "Full-time", "role": "Engineering", "role_type": "Devops", "salary_range": "$90 - $110", "equity_range": "", "experience": "Any (new grads ok)", "visa": "Will sponsor", "skills": [ "Python", "Docker" ] }, { "id": 63297, "title": "Firmware Engineer", "url": "https://www.ycombinator.com/companies/botbuilt/jobs/eH1oB3O-firmware-engineer", "location": "Durham, NC, US", "type": "Full-time", "role": "Engineering", "role_type": "Embedded systems", "salary_range": "$80K - $110K", "equity_range": "", "experience": "1+ years", "visa": "Will sponsor", "skills": [ "C++", "Distributed Systems", "Docker", "Device Drivers" ] }, { "id": 63298, "title": "Computer Vision and Optimization Engineer", "url": "https://www.ycombinator.com/companies/botbuilt/jobs/Joi1bpi-computer-vision-and-optimization-engineer", "location": "Durham, NC, US", "type": "Full-time", "role": "Engineering", "role_type": "Machine learning", "salary_range": "$90K - $130K", "equity_range": "", "experience": "1+ years", "visa": "Will sponsor", "skills": [ "Computer Vision", "Machine learning" ] }, { "id": 63514, "title": "Mechatronics Engineer", "url": "https://www.ycombinator.com/companies/botbuilt/jobs/thWY5Ph-mechatronics-engineer", "location": "Durham, NC, US", "type": "Full-time", "role": "Engineering", "role_type": "Embedded systems", "salary_range": "$90K - $120K", "equity_range": "", "experience": "1+ years", "visa": "Will sponsor", "skills": [ "C++", "Distributed Systems", "Docker", "Device Drivers" ] } ], "news": [ { "title": "Durham’s BotBuilt Using Robotic Arms To Revolutionize Affordable Housing – GrepBeat", "url": "https://grepbeat.com/2022/12/08/durhams-botbuilt-using-robotic-arms-to-revolutionize-affordable-housing/", "date": "Dec 08, 2022" } ], "launches": [] }, { "id": 23226, "name": "Soraban", "slug": "soraban", "former_names": [], "small_logo_thumb_url": "https://bookface-images.s3.amazonaws.com/small_logos/125412ffd8bc4ba73f837a877437db742e0a6d80.png", "website": "https://www.soraban.com", "all_locations": "Tempe, AZ, USA", "long_description": "AI-native Tax workflow Platform for accounting firms from client request to delivery. We raised $18m from patient investors who are onboard to help us grow and serve tens of thousands of accounting firms. We're not in this market to grow fast and fail fast. We're in this to build something that'll last decades. ", "one_liner": "AI tax workflow platform for accounting firms ", "team_size": 50, "industry": "B2B", "subindustry": "B2B -> Finance and Accounting", "launched_at": 1627685643, "tags": [ "AI-Enhanced Learning", "Robotic Process Automation", "SaaS" ], "tags_highlighted": [], "top_company": false, "isHiring": true, "nonprofit": false, "batch": "Winter 2021", "status": "Active", "industries": [ "B2B", "Finance and Accounting" ], "regions": [ "United States of America", "America / Canada", "Remote", "Partly Remote" ], "stage": "Early", "app_video_public": false, "demo_day_video_public": false, "app_answers": null, "question_answers": null, "url": "https://www.ycombinator.com/companies/soraban", "api": "https://devasheeshg.github.io/yc-api/batches/winter-2021/soraban.json", "year_founded": 2021, "city": "Tempe", "country": "US", "linkedin_url": "https://www.linkedin.com/company/sorabansoftware", "x_url": "", "fb_url": null, "cb_url": null, "github_url": null, "logo_url": "https://bookface-images.s3.amazonaws.com/small_logos/125412ffd8bc4ba73f837a877437db742e0a6d80.png", "app_video_url": null, "dday_video_url": null, "company_photos": [], "primary_partner": { "name": "Gustaf Alstromer", "url": "https://www.ycombinator.com/people/gustaf-alstromer" }, "founders": [ { "user_id": 322032, "full_name": "Enoch Ko", "title": "Founder/CEO", "founder_bio": "Enoch is the founder/CEO at Soraban. Software engineer and tax accountant background. Started Soraban in 2021 to streamline client data intake process and manual data entry for accounting firms.", "is_active": true, "linkedin_url": "https://www.linkedin.com/in/koenoch/", "x_url": "", "avatar_thumb_url": "https://bookface-images.s3.us-west-2.amazonaws.com/avatars/ff241c2451196b77fce79498abbb951dbe96c6e8.jpg" } ], "jobs": [ { "id": 61280, "title": "Full-Stack Software Engineer - Discovery", "url": "https://www.ycombinator.com/companies/soraban/jobs/1qaL5ow-full-stack-software-engineer-discovery", "location": "Chandler, AZ, US", "type": "Full-time", "role": "Engineering", "role_type": "Full stack", "salary_range": "$140K - $230K", "equity_range": "0.05% - 0.15%", "experience": "6+ years", "visa": "US citizen/visa only", "skills": [ "JavaScript", "Python", "React", "Ruby on Rails", "TypeScript" ] }, { "id": 85517, "title": "Tax Accountant (CPA, EA, or >3 years of experience)", "url": "https://www.ycombinator.com/companies/soraban/jobs/pJ0mF8y-tax-accountant-cpa-ea-or-3-years-of-experience", "location": "Phoenix, AZ, US", "type": "Full-time", "role": "Product", "role_type": null, "salary_range": "$80K - $160K", "equity_range": "", "experience": "3+ years", "visa": "US citizen/visa only", "skills": [] }, { "id": 95465, "title": "Customer Success Manager ", "url": "https://www.ycombinator.com/companies/soraban/jobs/oPOpQ9Z-customer-success-manager", "location": "Chandler, AZ, US", "type": "Full-time", "role": "Operations", "role_type": null, "salary_range": "", "equity_range": "", "experience": "3+ years", "visa": "US citizen/visa only", "skills": [] }, { "id": 104501, "title": "Full-Stack Software Engineer - Readiness", "url": "https://www.ycombinator.com/companies/soraban/jobs/Rvx0c9Y-full-stack-software-engineer-readiness", "location": "Chandler, AZ, US", "type": "Full-time", "role": "Engineering", "role_type": "Full stack", "salary_range": "$160K - $215K", "equity_range": "", "experience": "6+ years", "visa": "US citizen/visa only", "skills": [] }, { "id": 107363, "title": "VP of Marketing", "url": "https://www.ycombinator.com/companies/soraban/jobs/YeXSeBa-vp-of-marketing", "location": "Chandler, AZ, US", "type": "Full-time", "role": "Marketing", "role_type": null, "salary_range": "", "equity_range": "", "experience": "6+ years", "visa": "US citizen/visa only", "skills": [] }, { "id": 107561, "title": "Lead Software Engineer - Connect", "url": "https://www.ycombinator.com/companies/soraban/jobs/f1MenTw-lead-software-engineer-connect", "location": "Chandler, AZ, US", "type": "Full-time", "role": "Engineering", "role_type": "Backend", "salary_range": "$140K - $230K", "equity_range": "", "experience": "6+ years", "visa": "US citizen/visa only", "skills": [ "Amazon Web Services (AWS)", "Python" ] }, { "id": 108194, "title": "Tax Accountant (CPA, EA, or >3 years of experience)", "url": "https://www.ycombinator.com/companies/soraban/jobs/9bHJJcB-tax-accountant-cpa-ea-or-3-years-of-experience", "location": "Phoenix, AZ, US / Austin, TX, US", "type": "Full-time", "role": "Sales", "role_type": null, "salary_range": "$100K - $180K", "equity_range": "", "experience": "3+ years", "visa": "US citizen/visa only", "skills": [] } ], "news": [], "launches": [] }, { "id": 24817, "name": "Bytebot", "slug": "bytebot", "former_names": [ "tantl", "SaveCmd", "Tantl" ], "small_logo_thumb_url": "https://bookface-images.s3.amazonaws.com/small_logos/98cd71fc795fe80387da92b38a105f7c275db52d.png", "website": "https://www.bytebot.ai/", "all_locations": "Boston, MA, USA; Remote", "long_description": "Bytebot simplifies and streamlines the development of computer use agents with a ready made container for desktop automation, including APIs for mouse movements, keyboard input, and screenshot capture. \r\n\r\nThe container enables developers to build robust, dynamic computer use agents that can be configured and run anywhere.", "one_liner": "Bytebot is the open source container for AI desktop agents.", "team_size": 3, "industry": "B2B", "subindustry": "B2B -> Engineering, Product and Design", "launched_at": 1628647063, "tags": [ "Generative AI", "Robotic Process Automation", "SaaS", "AI", "AI Assistant" ], "tags_highlighted": [], "top_company": false, "isHiring": false, "nonprofit": false, "batch": "Summer 2021", "status": "Acquired", "industries": [ "B2B", "Engineering, Product and Design" ], "regions": [ "United States of America", "America / Canada", "Remote", "Fully Remote" ], "stage": "Early", "app_video_public": false, "demo_day_video_public": false, "app_answers": null, "question_answers": null, "url": "https://www.ycombinator.com/companies/bytebot", "api": "https://devasheeshg.github.io/yc-api/batches/summer-2021/bytebot.json", "year_founded": 2021, "city": "Boston", "country": "US", "linkedin_url": "https://www.linkedin.com/company/bytebot-ai/", "x_url": "", "fb_url": null, "cb_url": null, "github_url": null, "logo_url": "https://bookface-images.s3.amazonaws.com/small_logos/98cd71fc795fe80387da92b38a105f7c275db52d.png", "app_video_url": null, "dday_video_url": null, "company_photos": [], "primary_partner": null, "founders": [ { "user_id": 3186, "full_name": "Carl Atupem", "title": "Founder", "founder_bio": "Founder, CEO at Bytebot\n", "is_active": true, "linkedin_url": "https://www.linkedin.com/in/carl-atupem-94873245/", "x_url": "", "avatar_thumb_url": "https://bookface-images.s3.us-west-2.amazonaws.com/avatars/d1bd41b13aeeacc95fe6ba041fb95af8c8650f0f.jpg" } ], "jobs": [], "news": [], "launches": [ { "id": 81738, "title": "Bytebot: The easiest way to get web data", "tagline": "AI-powered scraping automations that evolve with the sites you target. Get the data you need with just a URL and a prompt.", "body": "![uploaded image](/media/?type=post&id=81738&key=user_uploads/3186/9e3e30fb-3ddb-47e7-aabf-860697f5ffb5)\n\n**TL;DR**\n\nBytebot uses AI to handle web data extraction and form filling. Developers use us to save time and effort, writing robust web scraping and automation logic that is reusable across many websites. Get started [here](http://app.bytebot.ai/auth/sign-up), and book some time with us [here](https://calendly.com/bytebot/bytebot-overview?back=1&month=2024-03).\n\n**❌ The Problem**\n\nBrowser scraping and web automation can be incredibly time-consuming and complex, often requiring a detailed and deep understanding of the web elements involved. Traditional methods are slow to iterate and can be difficult to scale, especially for startups and large companies.\n\nToday, you might:\n\n* Spend hours building scripts, hunting through HTML for the right selectors/xpaths\n* Waste effort debugging and maintaining scripts that break after page layout changes\n* Struggle with captchas, IP address rotation\n\n**💫 The Solution**\n\nBytebot simplifies and streamlines browser automation by translating your natural language prompts into intricate web tasks, from data extracting to website monitoring to form filling.\n\nBenefits include: \n\n* Eliminating time spent hunting for selectors or xpaths, as you can just prompt what you would like to happen instead\n* Allowing you to design scripts with  prompts that are reusable across multiple websites or domains\n* Providing robust and self-healing scripts that can adapt to website layout changes or updates\n\n![uploaded image](/media/?type=post&id=81738&key=user_uploads/3186/cb84607a-1f46-48f4-ad7b-c1c5f61dda6d)\n\n**⚒️How It Works**\n\nBytebot lets developers focus more on the prompting and less on the maintenance. Instead of writing complex scripts, you simply describe what you want to do, like _\"Add the item to the shopping cart\"_ or “_Login in with these credentials_“ and Bytebot handles the rest.\n\n![uploaded image](/media/?type=post&id=81738&key=user_uploads/3186/5317da8e-9760-4477-9a23-45ea0db84bc8)\n\nBehind the scenes, Bytebot takes a snapshot of a given webpage and leverages LLMs to translate prompts into the appropriate browser actions.\n\n![uploaded image](/media/?type=post&id=81738&key=user_uploads/3186/27874219-8f4a-4fcd-8480-21d039d83be6)\n\n**Speeding Up Automations**\n\nBytebot significantly speeds up the development cycle. Faster iterations enable companies to test and experiment with workflows without lengthy setup times. This increase in efficiency leads to higher ROI, as teams can focus on high-impact projects instead of getting bogged down by repetitive tasks.\n\n\n\n**Versatile for All Users**\n\n Bytebot is designed for users of all technical levels:\n\n* **For Companies with Existing Infrastructure:** If your company already has a mature Puppeteer or Playwright setup, Bytebot integrates seamlessly. You can quickly hook Bytebot into your existing scripts without overhauling your current system.\n* **For Companies Needing Browser Support:** If you don't want to manage browser infrastructure, our Hosted Browser API has you covered. We'll handle running the browser, and you can interact with Bytebot without worrying about captchas and IP address rotation.\n*  **For Non-Technical Users:** For the quickest setup and experimentation, our Playground (coming soon!) offers a no-code UI, with a live browser view.\n\n**🙏Our Asks**\n\n* [Sign up](http://app.bytebot.ai/auth/sign-up) to try Bytebot! Our tiered [pricing](https://www.bytebot.ai/#pricing) has options for individual engineers and companies of all sizes.\n* Follow Us on [LinkedIn](https://www.linkedin.com/company/bytebot-ai) and [Twitter](https://twitter.com/bytebot_ai) for new features, updates, and industry insights.\n* Share Bytebot with anyone who is looking to streamline their web automation.\n* If you have a few minutes to chat about Bytebot, book a time [here](https://calendly.com/bytebot/bytebot-overview?back=1&month=2024-03).", "url": "https://www.ycombinator.com/launches/LGM-bytebot-the-easiest-way-to-get-web-data", "votes": 61, "created_at": "2024-06-25T18:02:20.114Z" } ] }, { "id": 25616, "name": "Stackshine", "slug": "stackshine", "former_names": [], "small_logo_thumb_url": "https://bookface-images.s3.amazonaws.com/small_logos/2f75e99dea11ca83cb5f3e5de76cf447b57d842e.png", "website": "https://www.stackshine.com", "all_locations": "Portland, OR, USA; Providence, RI, USA; Remote", "long_description": "Stackshine is creating mission control for enterprise IT teams. We discover all the software being used across their organization and then automate workflows related to onboarding/offboarding, cost savings, and security. ", "one_liner": "Automating all IT operations", "team_size": 7, "industry": "B2B", "subindustry": "B2B -> Operations", "launched_at": 1646144116, "tags": [ "Robotic Process Automation", "Productivity", "Analytics", "Enterprise", "Data Engineering" ], "tags_highlighted": [], "top_company": false, "isHiring": false, "nonprofit": false, "batch": "Winter 2022", "status": "Acquired", "industries": [ "B2B", "Operations" ], "regions": [ "United States of America", "America / Canada", "Remote", "Fully Remote" ], "stage": "Early", "app_video_public": false, "demo_day_video_public": false, "app_answers": null, "question_answers": null, "url": "https://www.ycombinator.com/companies/stackshine", "api": "https://devasheeshg.github.io/yc-api/batches/winter-2022/stackshine.json", "year_founded": 2020, "city": "Portland", "country": "US", "linkedin_url": "https://www.linkedin.com/company/stackshine", "x_url": "https://twitter.com/StackshineSaaS", "fb_url": null, "cb_url": "https://www.crunchbase.com/organization/stackshine", "github_url": null, "logo_url": "https://bookface-images.s3.amazonaws.com/small_logos/2f75e99dea11ca83cb5f3e5de76cf447b57d842e.png", "app_video_url": null, "dday_video_url": null, "company_photos": [], "primary_partner": null, "founders": [ { "user_id": 49962, "full_name": "Tyler Diaz", "title": "Founder", "founder_bio": "Builder of tech", "is_active": true, "linkedin_url": "https://www.linkedin.com/in/tylerjdiaz/", "x_url": "https://twitter.com/tylerdiaz", "avatar_thumb_url": "https://bookface-images.s3.us-west-2.amazonaws.com/avatars/dd7c36c437e2147eedbdd5d5e159787332924d8d.jpg" } ], "jobs": [], "news": [], "launches": [] }, { "id": 26902, "name": "Maya Labs", "slug": "maya-labs", "former_names": [], "small_logo_thumb_url": "https://bookface-images.s3.amazonaws.com/small_logos/93bc27dea8b2f9bdbb3b58f34271dadc324cdf6d.png", "website": "https://mayalabs.io", "all_locations": "Bengaluru, KA, India", "long_description": "Applied research lab on a mission to build machines which can program themselves to do any task.", "one_liner": "Building self-programming machines", "team_size": 5, "industry": "B2B", "subindustry": "B2B -> Productivity", "launched_at": 1657029012, "tags": [ "Generative AI", "Robotic Process Automation", "B2B", "AI" ], "tags_highlighted": [], "top_company": false, "isHiring": false, "nonprofit": false, "batch": "Summer 2022", "status": "Active", "industries": [ "B2B", "Productivity" ], "regions": [ "India", "South Asia" ], "stage": "Early", "app_video_public": false, "demo_day_video_public": false, "app_answers": null, "question_answers": null, "url": "https://www.ycombinator.com/companies/maya-labs", "api": "https://devasheeshg.github.io/yc-api/batches/summer-2022/maya-labs.json", "year_founded": 2020, "city": "Bengaluru", "country": "IN", "linkedin_url": "", "x_url": "https://twitter.com/MayaLabsIO", "fb_url": null, "cb_url": null, "github_url": "https://github.com/mayahq", "logo_url": "https://bookface-images.s3.amazonaws.com/small_logos/93bc27dea8b2f9bdbb3b58f34271dadc324cdf6d.png", "app_video_url": null, "dday_video_url": null, "company_photos": [], "primary_partner": null, "founders": [ { "user_id": 272578, "full_name": "Sibesh Kar", "title": "Founder", "founder_bio": "Trying to build machines which can program themselves to do any task. ", "is_active": true, "linkedin_url": "https://www.linkedin.com/in/sibeshkar/", "x_url": "https://twitter.com/sibeshkar", "avatar_thumb_url": "https://bookface-images.s3.us-west-2.amazonaws.com/avatars/c196e49d6285d159a9eb2938f86cb5560de01c69.jpg" }, { "user_id": 276514, "full_name": "Shubham Mishra", "title": "Founder", "founder_bio": "Reliability, Server infrastructure, networking, distributed systems and everything else at Maya Labs. \nEx Ford Mobility India (Infra and Product). Ex Hyperloop India University Team.", "is_active": true, "linkedin_url": "https://www.linkedin.com/in/shbhmshr/", "x_url": "https://twitter.com/shbhmmshr", "avatar_thumb_url": "https://bookface-images.s3.us-west-2.amazonaws.com/avatars/a40bcc8b092d6ea5c37aebf1b9155ddf508be9c8.jpg" } ], "jobs": [], "news": [ { "title": "7 AI startups that stood out in YC's Summer '22 batch", "url": "https://techcrunch.com/2022/09/08/7-ai-startups-that-stood-out-in-ycs-summer-22-batch/?tpcc=tcplustwitter", "date": "Sep 08, 2022" } ], "launches": [ { "id": 65194, "title": "🛠️ 🧠 Maya Labs : Build custom software in minutes, using natural language", "tagline": "We're building a new way to program a machine using just natural language - starting off with a fast way to build internal tools by describing them in English.", "body": "Hi, we are Sibesh and Shubham, founders of Maya Labs. We are on a mission to teach machines how to code - starting off with a fast way to build internal tools in minutes using just natural language. Here’s an early look :\n\n![uploaded image](/media/?type=post&id=65194&key=user_uploads/272578/a3e654bf-56a0-48e4-abeb-8922bc23b8b0)\n\nA longer 5-min demo [here](https://www.youtube.com/watch?v=NPm0idKktzs) :\n\nToday, you can build one particular type of custom app well using Maya that let's you :\n\n1. **Fetch custom data** : Configure setup + query custom data from an external source (currently we do SQL, Gsheets, Notion, Airtable etc, but are expanding fast)\n2. **Visualize it** : Plot and visualize it in a custom table, line chart or bar graph\n3. **Perform actions** : Select and act on a particular data item (send a custom email, upload a file, trigger an API, update a DB item)\n\nMaya incrementally generates programs & shows results in response to steps you type in English, so the whole experience doesn’t feel like debugging someone else’s code. It’s possible to modify, delete & debug the natural language parts and Maya responds instantly - which makes it possible to hold and move around the _natural language program_ in your head. Checkout an early version of the docs [here](https://mayalabs.io/docs). We intend to keep it just one page long.\n\nThere’s still a lot to figure out and we’re trying to nail the experience for this one narrow use case first. When this works, we’ll try to extend it to just about any task you can think of - web navigation, connecting APIs, platform robots, RPA, home automation. Here’s a [show reel](https://mayalabs.io/library) of what else you can build manually with Maya today.\n\n**Who is this for?**\n\nMost internal tool builders are used by developers to build software for others on the product/ops/growth/sales/data teams, but we think this is a way for the user of the tool to also be the creator.\n\n**How does this work?**\n\nWe use a combination of conditional logic, neural search/classification, fine-tuned language models and template generation to make this work **predictably** and **reliably** in response to words you type/modify. We generate programs in a custom flow-based programming language that lets us express a wide range of behaviours with a smaller program length and visual DAGs. We also created our own benchmark to measure how well machines are at generating code (see our paper on this [here](https://blogs.mayalabs.io/benchmark)).\n\n**About us** :\n\nWe met in college many years back while starting Hyperloop India, the first Indian team at the SpaceX Hyperloop Competition - and have worked on-and-off on various projects since, sharing a common interest for hard engineering problems. Program synthesis is the hardest of them all - if software is eating the world, and you could build software that could build other software, why work on anything else?\n\nWe’ve been tinkering on this for years. Many Python libraries, papers, toy environments and product iterations later, we think we finally have an angle of attack that has a chance of working.\n\n**Ask** :\n\n1. Help us test our system! If you’d like to play around with or think someone in your team might benefit - send me a mail at sibesh\\[at\\]mayalabs\\[dot\\]io.\n2. Improve Maya with useful data - Let me know if you think a particular kind of program pattern we could teach Maya would be useful for you and a wide variety of other similar users.\n3. Introduce us to product/ops/growth/sales/data teams in your org who would like to build custom internal tools themselves without engineering effort\n4. Hit me up to learn more about our approach! Program synthesis, robot-run economies, programs that build other programs (including the YCombinator function haha)", "url": "https://www.ycombinator.com/launches/GxW-maya-labs-build-custom-software-in-minutes-using-natural-language", "votes": 53, "created_at": "2022-08-08T10:17:18.639Z" } ] }, { "id": 27255, "name": "Intuned", "slug": "intuned", "former_names": [ "The Metrics Shop", "Metrics Shop" ], "small_logo_thumb_url": "https://bookface-images.s3.amazonaws.com/small_logos/86c20a802b3a2e0667f7dcdc7fa4c4cb22913691.png", "website": "https://intunedhq.com", "all_locations": "Seattle, WA, USA", "long_description": "Intuned is a code-first browser automation platform with an AI agent that builds and maintains your automations. Developers use Intuned to deploy scrapers, RPAs, and crawlers as production-ready TypeScript or Python projects — with stealth, CAPTCHA solving, proxies, auth, and scheduling handled out of the box. The agent builds projects from a prompt, fixes failed runs, and can self-heal automations autonomously when websites change.", "one_liner": "Code-first scrapers and RPAs — built and maintained by AI", "team_size": 15, "industry": "B2B", "subindustry": "B2B -> Engineering, Product and Design", "launched_at": 1657233376, "tags": [ "Developer Tools", "Robotic Process Automation", "B2B", "API", "AI" ], "tags_highlighted": [], "top_company": false, "isHiring": false, "nonprofit": false, "batch": "Summer 2022", "status": "Active", "industries": [ "B2B", "Engineering, Product and Design" ], "regions": [ "United States of America", "America / Canada", "Remote", "Partly Remote" ], "stage": "Early", "app_video_public": false, "demo_day_video_public": false, "app_answers": null, "question_answers": null, "url": "https://www.ycombinator.com/companies/intuned", "api": "https://devasheeshg.github.io/yc-api/batches/summer-2022/intuned.json", "year_founded": 2022, "city": "Seattle", "country": "US", "linkedin_url": "https://www.linkedin.com/company/82287935", "x_url": "https://twitter.com/IntunedHQ", "fb_url": "https://www.facebook.com/IntunedHQ/", "cb_url": null, "github_url": null, "logo_url": "https://bookface-images.s3.amazonaws.com/small_logos/86c20a802b3a2e0667f7dcdc7fa4c4cb22913691.png", "app_video_url": null, "dday_video_url": null, "company_photos": [], "primary_partner": null, "founders": [ { "user_id": 224423, "full_name": "Faisal Ilaiwi", "title": "Founder", "founder_bio": "I am a software engineer. I started my career at Microsoft (6.5 years) and then joined UiPath (3.5 years). In UiPath, I had the roles of an Engineering Manager and a Director of Engineering, where I grew my team from 2 to ~20 people within 2.5 years and faced many challenges doing that. We are building Intuned to help ease some of these challenges.\n\nI grow up in Palestine and went to college there; I was then hired to join Microsoft from there.", "is_active": true, "linkedin_url": "https://www.linkedin.com/in/faisal-ilaiwi-77356135/", "x_url": "https://twitter.com/faisalilaiwi", "avatar_thumb_url": "https://bookface-images.s3.us-west-2.amazonaws.com/avatars/c9d0cd0dc50914ae87de900b1688042a650eb194.jpg" } ], "jobs": [], "news": [], "launches": [ { "id": 106701, "title": "WebSkillet.ai - web automation that gets cheaper every run", "tagline": "One API to automate any website. The agent writes code on run 1 and reuses it on every run after - script speed, agent resilience. Try it free at webskillet.ai", "body": "**Launch Video: [https://youtu.be/mXxe3P7hMzg](https://youtu.be/mXxe3P7hMzg)**\n\n**\\\nWhat is [Webskillet.ai](http://Webskillet.ai)?**\n\nWebskillet is a single API for web automation - extracting data, downloading files, crawling sites, filling forms, research tasks. You send a task in plain English (plus optional parameters, output schema, and start URL), and an agent executes it in a real browser using code.\n\nThat code is saved as a skillet, and since most web automation is repeat work (the same task on a schedule, or the same task with different parameters), the skillet gets reused on every later run. That's why run 2 can be 4x faster and cheaper than run 1. Here are [**20 real tasks with run 1 vs. run 2 numbers**](https://docs.google.com/spreadsheets/d/1EQVW4sypjW-v-sOz2bFU_iwYP3dadeA-4hdCYPYnHy0/edit?usp=sharing).\n\n**Why this is different?**\n\nPure AI browser agents treat every run like the first one. Nothing transfers: the agent re-explores, re-reasons, and you pay full price in time and tokens every time. Fine for one-offs, but web automation is mostly repeat work.\n\nWebskillet captures what the agent learns as code in the skillet, so run 2 starts from a working solution instead of a blank browser. And it's code reuse, not caching: same skillet, new parameters, fresh data from the live site.\n\n![uploaded image](/media/?type=post&id=106701&key=user_uploads/224423/bb695aa0-402f-4ded-9b33-c9c5ba7da222)\n\n**Our story**\n\nWe've been building [**Intuned**](http://intunedhq.com/) for the past 3 years: a browser automation platform that's been powerful for our customers, but it takes some setup to get up and running. Meanwhile, the world is moving to a place where agents just want everything as an API.\n\nWebskillet is our new product: everything we learned, packaged differently. One endpoint, no setup, and our infra (stealth browser, proxies, captcha solving, AI agent) behind it.\n\nDemo: [**https://youtu.be/U9WPX2bG_GI?si=0qoXtt7JKaTP-eXG**](https://youtu.be/U9WPX2bG_GI?si=0qoXtt7JKaTP-eXG)\n\n**Who it's for**\n\n* **AI agent products** that need the web as a tool. An example: a sales agent calling Webskillet to scrape a big website and build target lists\n* **Product teams** whose product needs data from sites that don't have an API in a repeat manner: government sites or legacy systems\n* **Growth/GTM engineers** doing enrichment and competitor monitoring: pricing pages, job postings, hiring signals\n* **Ops teams** doing repeated lookups: insurance verification, compliance checks, order status\n* **Data/analyst teams** building on web data: price intel, listings, market data, 10-K filings\n\n[**Here's a sheet of 20 real tasks we ran**](https://docs.google.com/spreadsheets/d/1EQVW4sypjW-v-sOz2bFU_iwYP3dadeA-4hdCYPYnHy0/edit?usp=sharing), with costs and timings for run 1 vs. run 2. Every row has a \"try it\" link so you can run yourself.\n\n**Asks**\n\n* **Try it! Sign up at [webskillet.ai](http://webskillet.ai)**\n* Watch the demo and tell us what you think, especially on the skillet concept\n* If you has a scraping/automation [**use case**](https://cal.com/team/intuned/quick-chat), lets talk!", "url": "https://www.ycombinator.com/launches/Rkz-webskillet-ai-web-automation-that-gets-cheaper-every-run", "votes": 23, "created_at": "2026-07-21T14:13:11.426Z" }, { "id": 99778, "title": "Intuned - Code-first browser automation, built and maintained by AI", "tagline": "We help developers build and maintain browser automations (scrapers, RPAs, crawlers) with an AI agent that writes the code and fixes it when websites change.", "body": "Hey Everyone,\n\nI'm Faisal, co-founder of Intuned. We're a code-first browser automation platform with an AI agent that builds and maintains your automations: scrapers, RPAs, crawlers, and anything else that needs to control a browser at scale.\n\n\n\n### The problem\n\nBrowser automations run a surprising amount of the internet's back office: price monitoring, job aggregators, data migrations, every RPA that exists because a vendor won't ship an API. And they're miserable to own. The scripts are fragile, selectors break constantly, and when a site changes, someone's day is gone.\n\nThat's before the infrastructure: stealth, proxies, CAPTCHAs, authentication, scheduling, retries, observability. Building one automation is annoying. Running a hundred in production is a full-time team.\n\nRecent AI browsers and LLM-driven agents handle one-off tasks well, but for repetitive tasks that run regularly, you need the speed, predictability, and cost profile that only code gives you. Nobody wants to write that code. And nobody wants to maintain it.\n\n### How Intuned works\n\nTwo parts, designed to work together.\n\n**The platform.** Write automations in TypeScript or Python, organize them into Projects, deploy to Intuned. Run them on demand via API or on a schedule. We handle stealth, CAPTCHA solving, proxies, AuthSessions for login flows, batched Jobs with retries and concurrency, and full observability (logs, browser traces, session recordings).\n\n![uploaded image](/media/?type=post&id=99778&key=user_uploads/224423/e2c5bb13-ee8a-497c-8312-c0793cad9183)\n\n**Intuned Agent.** Describe what you want. The agent explores the site, proposes a schema, writes the code, validates it against the live site, and delivers a production-ready Project you can review and deploy. It also edits existing projects and diagnoses failed runs by pulling real traces and logs, the same signals a human engineer would use.[ Learn more about it in our docs.](https://intunedhq.com/docs/main/02-intuned-agent/overview)\n\n**Self-healing projects.** Flip a switch and the project becomes autonomous. When a site changes and a run fails, Intuned detects it, the agent diagnoses and fixes it, and the project redeploys. You configure how much autonomy it gets. [learn more about it in our docs](https://intunedhq.com/docs/main/02-intuned-agent/self-healing-projects).\n\n**More demos:** [Build a scraper](https://www.youtube.com/watch?v=Es390qwW87w) - [Build an RPA](https://www.youtube.com/watch?v=D1Uf42OS1cs) - [Fix with AI](https://www.youtube.com/watch?v=dPqyOy-brOE) - [self-healing projects](https://youtu.be/22IYRA-ZqQk)\n\n### Use cases\n\n* **Scrapers** — pricing, job listings, public records, news, product catalogs. Anything structured you need on a schedule.\n* **RPA** — form submissions, data entry, report downloads, account actions. Essential when the service you need to integrate with doesn't have an API.\n* **Migrations** — moving data between services that don't talk to each other\n* **Crawlers** — multi-page discovery, sitemaps, content indexing at scale\n\n### How we got here\n\nWe've been in browser automation for over two years. We launched Intuned as a developer platform, but the #1 ask from customers was for us to just build and maintain the automations for them. We ended up running a services motion alongside the product, and over the past year we've processed **40M automation runs (20M minutes)** for a limited set of customers. It became one of the most loved parts of working with us, and also the thing we most wanted to replace with product.\n\nWe spent the last year trying, and nothing we built was good enough.\n\n**Then we integrated the Claude Agent SDK (Claude code wrapped as SDK) directly into Intuned.** Combined with the platform we'd already built, the agent now does what a human engineer on our team used to do: build projects, validate against the live site, read failed run traces, and ship fixes.\n\n### Our ask\n\n* Checkout our docs [here](https://intunedhq.com/docs/main/00-getting-started/introduction).\n* Try it free: [app.intuned.io](https://app.intuned.io)\n* Have a use case you want to talk through? [Book a demo](https://cal.com/team/intuned/get-started) or email [support@intunedhq.com](https://mailto:support@intunedhq.com)\n* Share it with anyone you think will be interested.\n\nThanks for reading 🙏\n\nFaisal & the Intuned team", "url": "https://www.ycombinator.com/launches/PxK-intuned-code-first-browser-automation-built-and-maintained-by-ai", "votes": 67, "created_at": "2026-04-15T14:35:16.537Z" }, { "id": 83938, "title": "Intuned Auth Sessions - Build authenticated scrapers and RPA", "tagline": "Intuned Auth Sessions enable you to easily build authenticated scrapers and browser automation for a wide set of use cases. Watch our demo videos to learn more!", "body": "# **The Problem**\n\nAdding authentication to any type of browser automation complicates it. There is a lot to handle—session expiration, proxies, obtaining user credentials/consent, monitoring, and more—and there is no easy or standard way to handle it.\n\n# **The Solution**\n\nIntuned Auth Sessions provides a standardized, developer-friendly way to add authentication to any browser automation task. Whether you're building authenticated scrapers, implementing RPA processes, or creating custom integrations for a service without APIs, Intuned simplifies the process and provides you with the infrastructure. Intuned’s auth sessions come with features that handle common authentication challenges, such as securing credentials, solving challenges, proxy support, 2fa, and more!\n\nOne of the features that we are super excited about is our auth session recorder, where you can allow the user to log in to a remote browser and capture the session from that. Check out how it works here (and watch the full video below for more info):\n\n![uploaded image](/media/?type=post&id=83938&key=user_uploads/224423/47827269-e4ce-4a87-b57b-7a07d8cdc9bd)\n\n# **Use Cases, Demos, and Docs**\n\n* **Build a LinkedIn integration/Build an integration with a service that has no APIs.**\n * Video Demo: \n * What does the demo show? We build an integration that enables users to “Connect to LinkedIn” and then show how we can automate a simple action.\n* **Authenticated Scraper:** Build scrapers that pull data from a website that requires auth.\n * Video Demo: \n * What does the demo show? We build a scraper that runs daily and pulls data from an authenticated website.\n* **Back office Automation:** Automate back office operations on a website that requires auth.\n * Video Demo: \n * What does the demo show? We automate some tasks on OrangeHRM using a service account.\n* To learn more about the feature and about Intuned in general, here are a few pointers to our docs:\n * General docs: \n * Auth Session docs: \n * How to guides: [how to automate LinkedIn](https://docs.intunedhq.com/docs/guides/auth/how-to-automate-linkedin) and [how to do credentials based authentication](https://docs.intunedhq.com/docs/guides/auth/how-to-authenticate-with-credentials).\n\n# **Our Ask & Onboarding**\n\n* Does this set of features sound like something you need? We are ready to onboard developers starting today: \n* Do you know someone facing these challenges? [Email us](mailto:founders@intunedhq.com)", "url": "https://www.ycombinator.com/launches/Lpq-intuned-auth-sessions-build-authenticated-scrapers-and-rpa", "votes": 30, "created_at": "2024-09-11T14:20:22.252Z" }, { "id": 81730, "title": "Intuned - The browser automation platform for developers and product teams", "tagline": "Develop, deploy, and monitor reliable web scrapers and browser automations", "body": "### **TL;DR**\n\nUse [Intuned](https://intunedhq.com/) to develop, deploy, consume, and monitor reliable web scrapers and browser automations. Reliability is powered by AI.\n\n### **❌ The Problem**\n\nStartups increasingly rely on browser automation for data scraping and RPA (Robotic Process Automation: automate what doesn’t have APIs via UI). However, the process of building, deploying, and maintaining these automations is fraught with challenges:\n\n* Building and deploying browser automations requires specialized infrastructure that can be complex to set up and maintain.\n* Browser-based automations are inherently fragile. Over 100 conversations with founders, product managers, and integration team members revealed a unanimous concern: the reliability and maintenance headache of scrapers and RPA.\n* The lack of monitoring tools for browser automations complicates the prompt detection and resolution of issues that occur during automation operations.\n\n### **✨ Our solution**\n\n[Intuned](https://docs.intunedhq.com/) is the platform for developers and product teams to develop, deploy, consume, and monitor **reliable** web scrapers and browser automations. We use LLMs to make common browser automation tasks easier to build and more reliable. Check out our docs ([https://docs.intunedhq.com](https://docs.intunedhq.com/)) for more info.\n\n\n\n### **How does it work?**\n\n**Develop**\n\nThe Intuned platform has an IDE built to speed up the process of building browser automations. Intuned IDE comes with custom features like action recording, selector creator, and more! To learn more, check out [**Develop**](https://docs.intunedhq.com/docs/platform/develop/overview).\n\n![uploaded image](/media/?type=post&id=81730&key=user_uploads/1063634/fd5e1fb7-0bff-44be-9444-1c33f8f14e84)\n\n**Deploy**\n\nIntuned’s projects can be deployed with a single click and within a minute. To learn more, check out [**Deploy**](https://docs.intunedhq.com/docs/platform/deploy).\n\n![uploaded image](/media/?type=post&id=81730&key=user_uploads/1063634/7f553275-4f64-4325-a19a-87cb7dd6b3d8)\n\n**Consume**\n\nStart consuming your projects by making API calls directly or scheduling jobs for scraping tasks. To learn more, check out [**Consume.**](https://docs.intunedhq.com/docs/platform/consume)\n\n![uploaded image](/media/?type=post&id=81730&key=user_uploads/1063634/0ac65a69-0a97-4777-b0b8-47ea18aa1ec0)\n\n**Monitor**\n\nGet proactive alerts on failures - with access to previous runs’ traces, recordings, and results. To learn more, check out [**Monitor.**](https://docs.intunedhq.com/docs/platform/monitor)\n\n![uploaded image](/media/?type=post&id=81730&key=user_uploads/1063634/3d319b39-591b-4947-97b7-0f086b324101)\n\n**Reliability**\n\nAs part of the Intuned platform, we provide the `@intuned/sdk` , a library built on top of Playwright, extending its capabilities by offering powerful helpers designed for common browser automation tasks. The `@intuned/sdk` enhances Playwright by making automation tasks more reliable and easier to implement.\n\n### **Our story - why build this?**\n\nFaisal and I started our YC journey working on an analytics platform. However, our background at UiPath drew founders to us with challenges around scaling manual processes. These interactions revealed the inefficiency of classic RPA tools for customer-facing integrations. When the time to pivot arrived, we analyzed YC startups batch-by-batch and observed a rising trend: many founders were venturing into areas requiring non-standardized access.\n\nOver 100 conversations with founders, product managers, and integration team members revealed a unanimous concern: the complexity and reliability of building and maintaining such integrations. Over the past few months, we've handled over a million requests for our early customers who are setting up their large-scale scrapers/automations.\n\n### **Our ask to you**\n\nAre you a dev building a scraper or browser automations? Or have you been postponing building those? Check out [Intuned](https://intunedhq.com/), or just chat with us [here](https://cal.com/team/intuned/get-started).\n\nKnow someone facing these challenges? We appreciate introductions! Contact us for an introduction blurb to facilitate the process.\n\n### **Docs and How-to guides**\n\nGeneral docs: \n\nQuick start: \n\nBuild your first scraper - webhook sink: \n\nBuild your first scraper - S3 sink: \n\nScrape data from web and files: \n\nWeb data extraction (how you don’t need selectors anymore!): ", "url": "https://www.ycombinator.com/launches/LGE-intuned-the-browser-automation-platform-for-developers-and-product-teams", "votes": 54, "created_at": "2024-06-25T15:00:44.188Z" }, { "id": 64781, "title": "🎵 Intuned - The data assistant for engineering leaders", "tagline": "We help engineering managers rein in the chaos", "body": "**TL;DR:** [Intuned](https://www.intuned.io/) is the data assistant for engineering leaders. [Watch our demo](https://bit.ly/3PIvCY2) to see how it works, and [email us](mailto:founders@intuned.io) if you want to learn more or get early access!\n\nWe are [Faisal](https://www.linkedin.com/in/faisal-ilaiwi-77356135/) and [Zawad,](https://www.linkedin.com/in/zawad-chowdhury/) and we are building [Intuned](https://www.intuned.io/)! We are on a mission to help engineering leaders!\n\n### **❌ The Problem**\n\nBeing an engineering manager day to day is not easy. There is a lot that engineering managers are dealing with, here are a few of the many issues we are trying to help with:\n\n* When it comes to people, many management decisions rely on gut feeling **alone.** This is why the performance-reviews period is a tough one.\n* There are a lot of repetitive, ad-hock, and manual tasks that are expected from a manager. An example is the regular reporting managers are often asked to prepare. Another one is the periodic planning activities that teams go through.\n* Driving change within organizations is hard, what makes it harder is the lack of an easy way to set measurable goals and track progress towards achieving them.\n* Managers have to deal with a set of different tools day to day, these tools all have different and inflexible notification systems, which makes them (as notification systems) useless. This is why managers use a pull model (go check on things) vs. a push model (they get notified) on the things they care about.\n\n## **✨ Our Solution**\n\nWhile the problems above are common with most managers, the exact way they can be solved will depend on the manager, their team, their processes, and the systems they use. There is NO one-size-fits-all here!\n\nWith [Intuned](https://www.intuned.io/)_,_ we are providing a toolbox to help!\n\nAs an engineering leader, [Intuned](https://www.intuned.io/) will enable you to connect your data from the different systems you and your team use (Jira, Github, Notion, Excel, and more) and will give you a powerful way to query it, explore it, build metrics, and actions on top of it.\n\nUsing this flexible approach, we believe we can help managers with many of their regular tasks, and save them time, effort and randomization.\n\n[Watch our 4-min demo to see how it works](https://www.youtube.com/watch?v=xzFTvSwFaqs)\n\n### **⚙️ How does it work?**\n\n**🔗 Connect your data with a few clicks.**\n\n![uploaded image](/media/?type=post&id=64781&key=user_uploads/224423/95454f4b-64db-4a06-9e08-b3550b138824)\n\n**❓ Query and explore your data - save hours on building manual reports.**\n\n![uploaded image](/media/?type=post&id=64781&key=user_uploads/224423/107f6c59-9790-4eb4-a040-78e18dffdeda)\n\n**🔔 Setup a custom notifications - never miss a critical issue again!**\n\n![uploaded image](/media/?type=post&id=64781&key=user_uploads/224423/482e12d8-d5c3-4eb8-be56-e78960e44a0d)\n\n🏗️ **Why build this?**\n\nAs a former engineering manager myself, this is the product that I wish I had!\n\n2\\.5 years ago, I became a manager for an engineering team of 2 that end up growing to around 20 people by the time I left _UiPath_.\n\nAs my team was scaling, we experienced growing pains with engineers being blocked, deployments getting delayed, and overall team execution lagging. I realized our software delivery process wasn’t keeping up but I couldn’t pin point what it was. So I ended up building a small tool to help! That tool ended up being very useful. That tool is the inspiration behind [Intuned](https://www.intuned.io/).\n\n### **Let’s Talk!**\n\nIf you or someone you know are a leader of an engineering team who is feeling some of these pains, we think [Intuned](https://www.intuned.io/) can help! please [email us](mailto:founders@intuned.io) or [schedule a meeting](https://calendly.com/intuned/15min) with us.", "url": "https://www.ycombinator.com/launches/Gqr-intuned-the-data-assistant-for-engineering-leaders", "votes": 241, "created_at": "2022-07-20T23:49:02.445Z" } ] }, { "id": 27882, "name": "Velos", "slug": "velos", "former_names": [ "GradientJ" ], "small_logo_thumb_url": "https://bookface-images.s3.amazonaws.com/small_logos/1521d60e05f3fa5d010d16b7722a46d4fb586f90.png", "website": "https://www.tryvelos.com", "all_locations": "San Francisco, CA, USA", "long_description": "Velos helps insurance carriers and MGAs automate complex, manual back-office tasks with AI workers instead of overseas teams.\r\n\r\nWe take over workflows with the same responsibility of a traditional overseas Business Process Outsourcing (BPO) firm: mapping the customer’s existing process, creating the SOP, building the automation, and continuously improving the workflow after launch. \r\n\r\nUnlike other \"automation platforms\" we don't rely on our customers to build and manage the workflow, we take over it for them like a traditional outsourcing firm.\r\n\r\nWe're automating the repetitive work people hate to do.", "one_liner": "AI BPO for Specialty Insurance MGAs", "team_size": 3, "industry": "Fintech", "subindustry": "Fintech -> Insurance", "launched_at": 1674503496, "tags": [ "Reinforcement Learning", "Robotic Process Automation", "Insurance", "Automation", "AI" ], "tags_highlighted": [], "top_company": false, "isHiring": false, "nonprofit": false, "batch": "Winter 2023", "status": "Active", "industries": [ "Fintech", "Insurance" ], "regions": [ "United States of America", "America / Canada" ], "stage": "Early", "app_video_public": false, "demo_day_video_public": false, "app_answers": null, "question_answers": null, "url": "https://www.ycombinator.com/companies/velos", "api": "https://devasheeshg.github.io/yc-api/batches/winter-2023/velos.json", "year_founded": 2023, "city": "San Francisco", "country": "US", "linkedin_url": "https://www.linkedin.com/company/tryvelos/", "x_url": "", "fb_url": null, "cb_url": null, "github_url": null, "logo_url": "https://bookface-images.s3.amazonaws.com/small_logos/1521d60e05f3fa5d010d16b7722a46d4fb586f90.png", "app_video_url": null, "dday_video_url": null, "company_photos": [], "primary_partner": { "name": "Nicolas Dessaigne", "url": "https://www.ycombinator.com/people/nicolas-dessaigne" }, "founders": [], "jobs": [], "news": [], "launches": [ { "id": 68762, "title": "GradientJ - Build and manage GPT3 🤖 applications at scale", "tagline": "GradientJ helps product teams deploy large language models at scale. Companies use us to create GPT3 applications and monitor them in production.", "body": "### **What is it?**\n\nGradientJ is a web application to build, compare, and deploy prompts for large language models like GPT3.\n\n### **Is this for me?**\n\nGradientJ is for:\n\n* Product teams that are just **thinking of using something like GPT3** in their product and are **trying to figure out how to get started**.\n* Teams who are **already using LLMs** to delight their customers and are **trying to level up their workflow** with faster iteration, better organization, easy cost and performance comparisons, and the on-ramp to \\~_the mythical land of fine-tuning\\~_.\n\n### **What’s the Problem?**\n\nCurrently, the **road from exciting prototype output with ChatGPT to production-ready GPT3 deployment is unclear** and littered with potential pitfalls.\n\nEven for the teams that do get version 1 out the door, **the workflow often ends up as a stitched together hodgepodge of makeshift processes and tools.**\n\nLLM deployment is a leading cause of “am I doing this right?”-itis. Symptoms include:\n\n* A text document with examples copy-pasted into the OpenAI playground\n* A prompt comparison process that consists of an eyeball test followed by “looks better-ish”\n* Hours spent tweaking a prompt on one example only to find, in the end, it breaks on all the others\n* Nightmares of users getting nonsensical or offensive output from a hallucinating LLM\n\n### **How do we solve it?**\n\nWe made a prompt engineering interface that:\n\nAutomatically suggests starting points or improvements to your prompts.\n\n![uploaded image](/media/?type=post&id=68762&key=user_uploads/314229/aabfc294-5d59-4f02-a9a0-ff9ea4e2a1b0)\n\nTracks prompt versions and benchmark examples so you can always compare and deploy the best iteration.\n\n![uploaded image](/media/?type=post&id=68762&key=user_uploads/314229/2491d923-0562-4c55-aa1c-fe0837a3b899)\n\nDeploys to your own API endpoint with automatic cost and performance tracking in one click.\n\n![uploaded image](/media/?type=post&id=68762&key=user_uploads/314229/df784795-2290-4ce9-9fbe-90ec41fbb21b)\n\n# **The Ask:**\n\n* **Let us help:** If you’re trying to use LLMs like GPT3 in your application and want to discuss how to **get version 1 out the door** **FAST**, we’d love to talk.\n* **Try it out:** If you’re already using LLMs in your application, we’d love to talk.\n* **Spread the Word:** If anyone you know is using or thinking of using LLMs, we want to turn their ideas into reality.\n\nDrop your email here , email us at [founders@gradientj.com](mailto:founders@gradientj.com), or jump right in and book a time to talk [here](https://calendly.com/daniel-gradientj/meeting)!", "url": "https://www.ycombinator.com/launches/Ht4-gradientj-build-and-manage-gpt3-applications-at-scale", "votes": 67, "created_at": "2023-01-26T20:03:07.818Z" } ] }, { "id": 28414, "name": "Automat", "slug": "automat", "former_names": [ "Lasso", "Automat (formerly lasso)" ], "small_logo_thumb_url": "https://bookface-images.s3.amazonaws.com/small_logos/55c324c494df9a6e5d939b7ecdc8e94880f2d575.png", "website": "https://www.runautomat.com/", "all_locations": "San Francisco, CA, USA", "long_description": "Automat is an AI automation company that replaces slow manual work and brittle legacy RPA with agents that understand context, recover from errors, and improve over time. Backed by a forward-deployed engineering team and enterprise-grade security (SOC 2, HIPAA, GDPR, ISO 27001), Automat runs reliable, human-in-the-loop automation for document-heavy industries — mortgage & lending, banking, insurance, healthcare, and manufacturing. It delivers this through two products: Automat Workforce and Automat Core.\r\n\r\nWorkforce - A digital teammate you onboard like a new hire. \r\nMeet Ace, a digital teammate with its own email, Slack, phone, and logins that works across the tools your team already uses. You reach it like a colleague (a text, a channel mention, an email, a doc comment) and it does the work end to end, asks when it needs judgment, and leaves an audit trail for everything it touches.\r\n\r\nLearn more here: https://runautomat.com/workforce \r\n\r\n\r\nCore - Managed agentic automation for enterprise \r\nManaged, AI-powered process automation for the enterprise. Automat Core replaces legacy RPA with self-healing agents that combine UI automation, AI document extraction (IDP), and API workflows in one platform. Automat's team builds, tests, and monitors every automation, so operations go live in days (not months), run reliably at scale, and cut operational costs — with human review wherever precision matters.\r\n\r\nLearn more here [https://runautomat.com/capabilities](https://runautomat.com/capabilities)", "one_liner": "Agentic automation that understands context, recovers, and improves.", "team_size": null, "industry": "B2B", "subindustry": "B2B -> Operations", "launched_at": 1677178978, "tags": [ "Documents", "Developer Tools", "Robotic Process Automation", "Automation", "AI" ], "tags_highlighted": [], "top_company": false, "isHiring": true, "nonprofit": false, "batch": "Winter 2023", "status": "Active", "industries": [ "B2B", "Operations" ], "regions": [ "United States of America", "America / Canada" ], "stage": "Growth", "app_video_public": false, "demo_day_video_public": false, "app_answers": null, "question_answers": null, "url": "https://www.ycombinator.com/companies/automat", "api": "https://devasheeshg.github.io/yc-api/batches/winter-2023/automat.json", "year_founded": 2023, "city": "San Francisco", "country": "US", "linkedin_url": "https://www.linkedin.com/company/90558949/admin/", "x_url": "", "fb_url": null, "cb_url": null, "github_url": null, "logo_url": "https://bookface-images.s3.amazonaws.com/small_logos/55c324c494df9a6e5d939b7ecdc8e94880f2d575.png", "app_video_url": null, "dday_video_url": null, "company_photos": [], "primary_partner": { "name": "Diana Hu", "url": "https://www.ycombinator.com/people/diana-hu" }, "founders": [ { "user_id": 415306, "full_name": "Lucas Ochoa", "title": "Founder", "founder_bio": "CEO at Automat | Ex:Google & Microsoft", "is_active": true, "linkedin_url": "https://www.linkedin.com/in/lucas-ochoa-51222678/", "x_url": "https://twitter.com/lucas0choa", "avatar_thumb_url": "https://bookface-images.s3.us-west-2.amazonaws.com/avatars/a3460bb02cae910a8167c7a514d4b7c6208771d6.jpg" }, { "user_id": 249452, "full_name": "Gautam Bose", "title": "Founder", "founder_bio": "Open Laptop, an Open Mind", "is_active": true, "linkedin_url": "https://www.linkedin.com/in/gautambose/", "x_url": "", "avatar_thumb_url": "https://bookface-images.s3.us-west-2.amazonaws.com/avatars/6cc3e7ee978641ccc9ae1932877dff2f101c8f69.jpg" } ], "jobs": [ { "id": 77831, "title": "Applied AI Engineer", "url": "https://www.ycombinator.com/companies/automat/jobs/JyD9j1f-applied-ai-engineer", "location": "San Francisco", "type": "Full-time", "role": "Engineering", "role_type": "Full stack", "salary_range": "$175K - $220K", "equity_range": "0.25% - 1.00%", "experience": "1+ years", "visa": "US citizen/visa only", "skills": [] }, { "id": 77834, "title": "Sales Development Representative", "url": "https://www.ycombinator.com/companies/automat/jobs/95CHPIW-sales-development-representative", "location": "San Francisco", "type": "Full-time", "role": "Sales", "role_type": null, "salary_range": "$90K - $130K", "equity_range": "0.02% - 0.05%", "experience": "Any (new grads ok)", "visa": "US citizen/visa only", "skills": [] }, { "id": 77835, "title": "Enterprise Account Executive (Full Cycle)", "url": "https://www.ycombinator.com/companies/automat/jobs/WDGBMTs-enterprise-account-executive-full-cycle", "location": "San Francisco", "type": "Full-time", "role": "Operations", "role_type": null, "salary_range": "$90K - $140K", "equity_range": "0.02% - 0.05%", "experience": "3+ years", "visa": "US citizen/visa only", "skills": [] }, { "id": 79159, "title": "Forward Deployed Engineer", "url": "https://www.ycombinator.com/companies/automat/jobs/QfHWU1w-forward-deployed-engineer", "location": "San Francisco", "type": "Full-time", "role": "Engineering", "role_type": "Full stack", "salary_range": "$145K - $185K", "equity_range": "0.05% - 0.20%", "experience": "3+ years", "visa": "US citizen/visa only", "skills": [] }, { "id": 81081, "title": "Senior Software Engineer", "url": "https://www.ycombinator.com/companies/automat/jobs/Uak8GDJ-senior-software-engineer", "location": "San Francisco", "type": "Full-time", "role": "Engineering", "role_type": "Full stack", "salary_range": "$175K - $220K", "equity_range": "0.05% - 0.25%", "experience": "3+ years", "visa": "US citizen/visa only", "skills": [] } ], "news": [ { "title": "Automat Raises $15.5M Series A | Automat", "url": "https://www.runautomat.com/blog/automat-raises-15-5m-series-a", "date": "Nov 19, 2025" }, { "title": "Y Combinator-backed Automat turns videos into workflow automations | TechCrunch", "url": "https://techcrunch.com/2023/09/06/y-combinator-backed-automat-turns-videos-into-workflow-automations/?guccounter=1&guce_referrer=aHR0cHM6Ly93d3cuZ29vZ2xlLmNvbS8&guce_referrer_sig=AQAAAMl7GrmIJwYOYFCoUcYfqNTrV0r82APvTlpVt0DMIEyjpjQS9MtJb9DFuiE2HtyaFOpOQzv4SZPUnB89Y9Lxdkdk1vXkAKBjYDjtkKkbSOpUGhc_GhcgUMy3aS5HO5L99OwXIn1t8_zFB9_NfDUPitSBQXBL-PWj4j52H5k2nzke", "date": "Sep 06, 2023" }, { "title": "These Y Combinator-backed startups are trying to build 'ChatGPT for X'", "url": "https://techcrunch.com/2023/04/04/these-y-combinator-startups-are-trying-to-build-chatgpt-for-x/", "date": "Apr 04, 2023" }, { "title": "The Top 34 Companies of the W23 Program with a focus on AI", "url": "https://www.cerebralvalley.com/post/a-synopsis-of-y-combinator-companies-in-the-w23-batch-of-startups", "date": "Mar 15, 2023" }, { "title": "AI-led change means businesses must adopt more quickly.", "url": "https://russmalz.com/ai-led-change-is-forcing-businesses-to-adapt-quickly-and-the-pace-is-accelerating/", "date": "Feb 26, 2023" } ], "launches": [ { "id": 117458, "title": "Ace from Automat: it's instinct, but for work", "tagline": "Hire an AI teammate with its own computer, phone number, inbox, and logins. It works inside the tools you already use and hands back finished work, not status updates.", "body": "Hi! Lucas here from Automat. Stoked to share what the team's been heads-down on: Ace. It's instinct, but for work. **The TL;DR: you can now hire an AI teammate.**\n\nEvery team we talk to wants AI teammates, but doesn't know how to actually put one to work. Maybe you've tried homebrewing OpenClaw or Hermes. The two usual options both whiff: traditional automation (UiPath) makes you sign up for a months-long integration project before it does one useful thing, and AI \"assistants\" hand you a wall of text, and you still go do the work yourself.\n\n**Ace is different.** You onboard it like any new hire, and it shows up where the work happens:\n\n• Forward it an email or cc its own inbox, and the draft lands in your Drive with the judgment calls left as comments.\n\n• @Ace in Slack after a call, and the summary + action items land on the opportunity.\n\n• Text it between meetings (\"move Thursday to Friday and tell them why\"), and it updates the invite, emails everyone, and confirms.\n\n• Ping it in a doc comment, or have it make a call on its own line.\n\nWhat comes back is finished work: CRM updated, email sent, invoice reconciled. When it hits a real fork in the road, it hands you the open questions instead of guessing.\n\n**Why trust it?** We spent two years automating high-volume work in mortgage, lending, and insurance, where a wrong answer is a compliance problem. Ace pairs a deterministic engine for must-be-identical work with AI for the fuzzy stuff. Every run is logged step by step, you set what needs approval, and you can revoke access like any employee's. SOC 2, encrypted credentials, role-based access.\n\n\"I really tried hard not to like this product, and I'm in love.\" — Asa Block, COO, Near Space Labs\n\n\"It's a no-brainer… Monday to Friday you won't believe how much more stuff you get done.\" — Jack Miller, President & CEO, T3 Sixty\n\n## Our asks 🙏\n\n• Try Ace: One month free with unlimited usage for the YC community only. Sign up here:() Use the promo code: ACEYC30\n\n• Intros: Know a team drowning in admin across a dozen systems? Send them our way.\n\n• We're hiring! ", "url": "https://www.ycombinator.com/launches/UYU-ace-from-automat-it-s-instinct-but-for-work", "votes": 1, "created_at": "2026-09-30T16:33:11.033Z" }, { "id": 69615, "title": "Automat RPA - Robotic process automation for Chrome using ChatGPT and computer vision", "tagline": "Automate anything in Chrome using natural language", "body": "### **TLDR: Traditional automation software is slow and expensive, Automat is 10X faster and 1/10th the price. It lets operations teams automate using natural language.**\n\nHowdy y’all! Lucas here from [Automat](https://www.getlassoai.com/) - excited to share what my cofounder Gautam and I have been working on!\n\n**🧐 The problem: Traditional Robotic Process Automation is slow and expensive.**\n\nIt’s extremely time consuming and expensive to build out robot process automations using traditional tools like UiPath (\\~$66,000 per yearly license - Yikes!) not to mention the initial consultant setup fees which most people are forced to pay if they don’t have the resources in house (at the lowest end we’re talking $50,000+ per project - Double Yikes!!)\n\n💡 **Our Solution: Automat makes it 10X cheaper and faster to build any robotic process automation using natural language on chrome.**\n\nWe’re solving this by creating a platform that enables robotics process automation for Chrome using large language models and computer vision.\n\nCurrently our customers send us descriptions or videos of the processes they’d like us to automate and we use our internal tooling to quickly build out those automations.\n\nHere is one of our customer’s attended bots in action, scraping an email, looking through a prospecting tool, and saving it to a spreadsheet.\n\n![uploaded image](https://bookface.ycombinator.com/media/?type=post&id=69615&key=user_uploads/415306/a547a684-faee-4c6e-b3db-88cef800940a)\n\n\\\nWe want to let anyone regardless of timeline or budget automate the work they do by using natural language or simply sending Automat the workflow in a screen recording.\n\n**👯 A little more about us**\n\n![uploaded image](/media/?type=post&id=69615&key=user_uploads/415306/e82023cd-a7c1-4be3-8446-f2e0c2eae6f6)\n\nGautam and Lucas met on the first day of Design school at Carnegie Mellon University and have been roommates ever since:\n\n* At Google, their team was tasked with finding product applications of [LaMDA](https://blog.google/technology/ai/lamda/) in early 2020, and they built over 50+ prototypes integrating LaMDA across Google’s products and services. Eventually Gautam would go on to implement multiple demos for Google’s [first public Demonstration of LaMDA](https://www.theverge.com/2022/5/11/23065072/google-ai-app-test-kitchen-future-io-2022), the AI test kitchen.\n* They were hired together at Google’s Creative Lab as part of the [Creative Lab 5](https://www.creativelab5.com/), at Google they spent the following three years launching products like the [AI Test Kitchen](https://blog.google/technology/ai/join-us-in-the-ai-test-kitchen/), [Everyday Helper Robots](https://everydayrobots.com/), [Teachable Machine](https://teachablemachine.withgoogle.com/), the [Google Pixel Buds Pro](https://store.google.com/product/pixel_buds_pro?hl=en-US&pli=1), and most importantly a [high speed lucky charms cereal sorter](https://coral.ai/projects/teachable-sorter) 🍀\n* At Carnegie Mellon they studied Human Computer Interaction and Design. During their Junior year they created an adorable [AI robotic assistant](https://vimeo.com/270823057) that caught Google’s attention which awarded them a grant during their senior year to [research Robotics and AI](https://vimeo.com/393172151).\n\n**🙏 Our asks**\n\nAre you struggling in a deluge of repetitive or complex tasks that you wish you could hire someone for (or automate) but don’t have the budget? Send an email to [lucas@runautomat.com](https://lucas@runautomat.com) or submit a request on our [website](https://www.getlassoai.com/).\n\nDoes the above sound like someone you know? We’d appreciate intros to anyone. Let us know and we can send you over a quick intro blurb to make it an easy email to forward.\n\n**YC Special Offer:** For the first 10 YC companies that reach out we’ll waive the setup fee associated with building your automation!", "url": "https://www.ycombinator.com/launches/I6p-automat-rpa-robotic-process-automation-for-chrome-using-chatgpt-and-computer-vision", "votes": 91, "created_at": "2023-02-23T19:08:21.664Z" } ] }, { "id": 28887, "name": "Skyvern", "slug": "skyvern", "former_names": [ "Wyvern AI (Previously Ikonomos)", "Wyvern AI" ], "small_logo_thumb_url": "https://bookface-images.s3.amazonaws.com/small_logos/87d977bf0d4b7032c71830901e118beb0ca9946c.png", "website": "https://www.skyvern.com/", "all_locations": "San Francisco, CA, USA; Remote", "long_description": "Skyvern helps companies automate browser based workflows using LLMs and Computer Vision, fully automating manual workflows and replacing brittle or unreliable scripts.\r\n\r\nExamples include:\r\n1. Automating materials procurement from commerce websites\r\n2. Completing complex multi-step workflows (ie getting an insurance quote from Geico.com)\r\n3. Automatically logging into portals and downloading invoices\r\n4. Navigating legacy content systems to do data extraction or data entry", "one_liner": "Open Source AI Agent to automate browser workflows via an API", "team_size": 11, "industry": "B2B", "subindustry": "B2B -> Infrastructure", "launched_at": 1687648021, "tags": [ "Robotic Process Automation", "Workflow Automation", "Open Source", "API", "AI" ], "tags_highlighted": [], "top_company": false, "isHiring": true, "nonprofit": false, "batch": "Summer 2023", "status": "Active", "industries": [ "B2B", "Infrastructure" ], "regions": [ "United States of America", "America / Canada", "Remote", "Fully Remote" ], "stage": "Early", "app_video_public": false, "demo_day_video_public": false, "app_answers": null, "question_answers": null, "url": "https://www.ycombinator.com/companies/skyvern", "api": "https://devasheeshg.github.io/yc-api/batches/summer-2023/skyvern.json", "year_founded": 2023, "city": "San Francisco", "country": "US", "linkedin_url": "https://www.linkedin.com/company/95726232", "x_url": "https://x.com/skyvernai", "fb_url": null, "cb_url": null, "github_url": null, "logo_url": "https://bookface-images.s3.amazonaws.com/small_logos/87d977bf0d4b7032c71830901e118beb0ca9946c.png", "app_video_url": null, "dday_video_url": null, "company_photos": [], "primary_partner": { "name": "Nicolas Dessaigne", "url": "https://www.ycombinator.com/people/nicolas-dessaigne" }, "founders": [ { "user_id": 1120458, "full_name": "Suchintan Singh", "title": "Founder", "founder_bio": "I'm the founder of Skyvern. We help companies automate complex workflows using AI agents and computer vision.\n\nIn the past, I've built the ML Platforms at both Faire and Gopuff, which generated over $100M of GMV across their two businesses, \n\nI also love bad jokes and puns, so if you have any or are in the need of any feel free to hit me up!", "is_active": true, "linkedin_url": "https://www.linkedin.com/in/suchintansingh/", "x_url": "https://x.com/suchintan", "avatar_thumb_url": "https://bookface-images.s3.us-west-2.amazonaws.com/avatars/e6e042c66b8d37bbdbefedec94f3d36e11cbdfda.jpg" }, { "user_id": 1157498, "full_name": "Shuchang Zheng", "title": "Founder", "founder_bio": "I am the CTO and cofounder of Skyvern. For the past 5 years, I've been building developer and platform tools.\n\nAt Lyft, I built testing tools (simulation platform, load test framework, and end to end test framework) used by 1000+ engineers to boost dev productivity, ensure Lyft doesn't go down during peak events, and reduce infra cost.\n\nAt Patreon, I helped scale the payment and db infra to process 20M+ transactions per month and free 100+ engineers from the 3-day monthly code freeze ", "is_active": true, "linkedin_url": "https://linkedin.com/in/shuchang-zheng-76784958", "x_url": "", "avatar_thumb_url": "https://bookface-images.s3.us-west-2.amazonaws.com/avatars/ff5b3572753bb2ab538c8c114dbd6cad084790f6.jpg" } ], "jobs": [ { "id": 76074, "title": "AI Agent Builder / Workflow Wizard", "url": "https://www.ycombinator.com/companies/skyvern/jobs/YFDIaHd-ai-agent-builder-workflow-wizard", "location": "CA / Remote (CA)", "type": "Full-time", "role": "Operations", "role_type": null, "salary_range": "$60K - $85K CAD", "equity_range": "0.05% - 0.10%", "experience": "Any (new grads ok)", "visa": "US citizen/visa only", "skills": [] }, { "id": 95748, "title": "Founding SDR, AI-Powered Outbound", "url": "https://www.ycombinator.com/companies/skyvern/jobs/3P7vhT3-founding-sdr-ai-powered-outbound", "location": "US / CA / Remote (US; CA)", "type": "Full-time", "role": "Sales", "role_type": null, "salary_range": "$60K - $75K", "equity_range": "0.01% - 0.10%", "experience": "Any (new grads ok)", "visa": "US citizen/visa only", "skills": [] }, { "id": 95749, "title": "Founding Account Executive - AI Infrastructure", "url": "https://www.ycombinator.com/companies/skyvern/jobs/9eucL4u-founding-account-executive-ai-infrastructure", "location": "US / CA / Remote (US; CA)", "type": "Full-time", "role": "Sales", "role_type": null, "salary_range": "$120K - $300K", "equity_range": "0.10% - 0.20%", "experience": "1+ years", "visa": "US citizen/visa only", "skills": [] }, { "id": 95750, "title": "Founding Developer Marketing – Open Source AI", "url": "https://www.ycombinator.com/companies/skyvern/jobs/1qRTlVx-founding-developer-marketing-open-source-ai", "location": "US / CA / Remote (US; CA)", "type": "Full-time", "role": "Marketing", "role_type": null, "salary_range": "$100K - $150K", "equity_range": "0.10% - 0.30%", "experience": "3+ years", "visa": "US citizen/visa only", "skills": [] } ], "news": [], "launches": [ { "id": 105883, "title": "Rustwright: Rewrite of Playwright in rust that uses 70% less memory (and is 2.55x faster)", "tagline": "Remove the subprocess and save so much memory", "body": "Hey everyone, Suchintan from Skyvern (YC S23). We run browser automation for AI agents at scale, and we're launching [Rustwright](https://github.com/Skyvern-AI/rustwright): [Playwright on an in-process Rust CDP engine.](https://github.com/Skyvern-AI/rustwright) The engine consumes 70% less memory than Playwright, and is 2.55x faster to boot.\n\n**We built this to improve Skyvern’s memory footprint**, and thought this would be independently valuable to the broader community\n\nThe whole idea fits in two lines:\n\n```\nplaywright-python: your code ──pipe──► Node driver (separate process) ──CDP──► Chromium\nrustwright: your code ────────────────── raw CDP ─────────────────────► Chromium\n```\n\nRemoving the driver changes three things:\n\n1. **Less abstraction.** The driver subprocess and its pipe speaks CDP straight to Chromium. In our runs, the new engine is 2.55× and **consumes 70% less memory** than the equivalent playwright process\n2. **No Playwright driver fingerprint.** The driver never loads, so its signatures never appear: no `__playwright__binding__` globals, no `Runtime.enable` on the default path (the well-known console-serialization leak). The claim is deliberately narrow: no Playwright-specific automation fingerprint, not \"undetectable\"\n3. **A single rust core**: a single Rust core backs both language bindings (Python with sync and async; Typescript) with plans to expand to many popular languages (Ruby / Go / Java to name a few). Within the supported surface, migration is often a one-line import change — `from rustwright.sync_api import sync_playwright.`Clicks and typing go through real CDP input events, not synthetic DOM calls, and cross-origin iframes auto-attach with `frame_locator()` routing across origins.\n\n[playwright-python](https://github.com/microsoft/playwright-python) pipes every call through a bundled Node driver process. [Rustwright](https://github.com/Skyvern-AI/rustwright) drives Chromium over raw Chrome DevTools Protocol from a Rust core, an async CDP client on Tokio (WebSocket, plus opt-in Unix-pipe transport), exposed in-process through thin PyO3 bindings for Python and napi-rs bindings for Node.\n\nStatus: alpha, Chromium-only\n\n**Want to give it a try? Change one line of code!**\n\n```jsx\npip install rustwright\n\n- from playwright.sync_api import sync_playwright\n+ from rustwright.sync_api import sync_playwright\n```\n\n⭐ Give it a try here and let us know what you think: ", "url": "https://www.ycombinator.com/launches/RXn-rustwright-rewrite-of-playwright-in-rust-that-uses-70-less-memory-and-is-2-55x-faster", "votes": 8, "created_at": "2026-07-15T17:10:22.738Z" }, { "id": 94724, "title": "Skyvern: Asking AI to build scrapers should be easy right?", "tagline": "Let AI maintain those pesky scrapers", "body": "![uploaded image](/media/?type=post&id=94724&key=user_uploads/1120458/6473b24c-3925-48bf-80c8-00430f81e5ef)\n\n**TL;DR - We just gave Skyvern the ability to write and maintain its own code, making it 2.7x cheaper and 2.3x faster. Give it a prompt (or a series of prompts), and the AI will generate and maintain playwright code while it runs. Try out the via [Open Source](https://github.com/Skyvern-AI/Skyvern?ref=skyvern.com) or [Cloud](http://app.skyvern.com/?ref=skyvern.com)**\n\n💡 **Recap: What is Skyvern?** It’s [an open source tool](https://github.com/Skyvern-AI/Skyvern?ref=skyvern.com) that helps companies automate things in the browser with AI. We use computer vision + LLMs to turn prompts into automations that run. We serve both technical and non-technical customers, and have helped them automate things like applying to jobs, fetching invoices or utility bills, filling out government forms, and purchase products from hundreds of different websites.\n\n# **Asking AI to build scrapers should be easy right? Right??**\n\n![](https://www.skyvern.com/blog/content/images/2025/10/image---2025-10-17T132106.418.png)\n\nSome of you may remember our [Hackernews launch](https://news.ycombinator.com/item?id=39706004&ref=skyvern.com) from last year. All of the discussion circled around the same idea: “Building the automation is the hard part… we just want Skyvern to write the code”.\n\nWe agreed. Keeping the agent in the loop means invoking an expensive and non-deterministic LLM call on every run. If Skyvern could compile its reasoning into code and run that instead of keeping an LLM in the loop, automations would become **faster, cheaper, and more reliable.**\n\nSo we tried to teach Skyvern to do exactly that… but it turns out, asking AI to write code the same way you and I would wasn’t easy. We ran into two big problems:\n\n1. Requirements for automations are ambiguous at best, and misleading at worst — even humans struggle to define them clearly\n2. The internet is messy: drop-downs masquerade as textboxes, checkboxes that are always checked, and search bars that are secretly buttons.\n\nGetting an agent to navigate that chaos, understand intent, and still produce maintainable code came through one major breakthrough: **reasoning models**.\n\nReasoning models unlock two important capabilities:\n\n1. They [boosted the agent accuracy](https://www.skyvern.com/blog/web-bench-a-new-way-to-compare-ai-browser-agents/) enough for production use\n2. They let the agent leverage its trajectory to write a script resembling something an engineer would write\n\n# **This is too abstract. When does this matter?**\n\nBefore we dive into the solution, let’s look at a real-world example: Registering new companies for payroll with [Delaware.gov](http://Delaware.gov)\n\n![](https://www.skyvern.com/blog/content/images/2025/10/image---2025-10-17T132100.688.png)\n\nHere’s a simple prompt that reliably powers the workflow:\n\n> _Your goal is to fill out the EIN registration form. Fill out the form until you're at the form confirmation page with a summary of all information. Your goal is complete once you see a summary of all of the information._\n\nein_info: {{ein_info}}\n\nWriting deterministic code should be easy right?\n\nHere’s what a naive AI Generated implementation looks like:\n\n![](https://www.skyvern.com/blog/content/images/2025/10/image-1.png)\n\nAnd here’s where it falls apart almost immediately.\n\n1. Coupled interactions. Choices on this form aren’t independent. Sometimes radio buttons are linked together, but aren’t represented as such in the DOM. Other times, different buttons trigger different follow-up questions, so a static script breaks as soon as you pick something unexpected.\n\n![](https://www.skyvern.com/blog/content/images/2025/10/image---2025-10-17T132055.697.png)\n\n**Random failures.** Government websites love to go down at night, change field layouts between sessions, or throw you a “try again later” page mid-run.\n\n![](https://www.skyvern.com/blog/content/images/2025/10/image---2025-10-17T132051.371.png)\n\n### **Coupled interactions**\n\nConsider the radio button example above. Any seasoned developer would know that these legal structures are linked together. You’d instinctively model them as a finite set of entity types, and create a tree-like script that branches into different paths based on the input.\n\nBut that abstraction doesn’t exist for an agent. To agents like Skyvern, this is just a list of buttons. The relationship between the users’ input and the available set of legal structures doesn’t exist ahead of time — it must be discovered at runtime.\n\nThe agent has to infer, from the DOM and the page transitions, which choices lead where.\n\n### **Random non-deterministic failures**\n\nAgents shine when things don’t go as planned. We don’t want to hard code every single edge case when compiling an agent into a deterministic script.. because we’re back to writing brittle scripts. Instead, we want to leverage agents for these situations.\n\nTake this government form: Delaware’s portal is unavailable at night or over the weekend. Or sometimes, it’ll require you to call the IRS or sending them a fax / mail to proceed with the form. You want some intelligence in the loop to handle these scenarios gracefully.\n\n# **So.. how can you codify this in an agent?**\n\nAfter a few runs like the ones above, we realized “have the agent write code” wasn’t enough. We needed to copy how developers actually work: figure out the flow, add logic where it breaks, and bake that behavior into Skyvern.\n\nSo we split its job into two:\n\n1. Explore mode, where the agent learns how to navigate a website for a given flow, generating any metadata necessary for it to operate in subsequent runs\n\nReplay mode, it compiles those learnings into deterministic Playwright and runs fast and cheap, only falling back to the agent when something new or weird happens\n\n![](https://www.skyvern.com/blog/content/images/2025/10/image---2025-10-17T132046.901.png)\n\n### **Explore once: get the agents’ trajectory**\n\nLet’s start with a plain prompt for Skyvern. The `ein_info` field is just a json blob with all of a company’s metadata (entity type, responsible party, etc). The goal of this explore run isn’t to finish fast, it’s to learn the flow and record a trajectory we can compile later.\n\n> _Go to [https://sa.www4.irs.gov/modiein/individual/index.jsp](https://sa.www4.irs.gov/modiein/individual/index.jsp)\\\n> and generate an SS4 with the following information: {{ ein_info }}_\n\n### **Step 1: generate a naive script**\n\nFrom that trajectory, the agent can spit out a basic Playwright script. It runs, but it’s brittle—no context, no fallbacks:\n\n![](https://www.skyvern.com/blog/content/images/2025/10/image-2.png)\n\n**What’s missing:** it doesn’t know _why_ it’s clicking “Corporation,” what should appear next, or how to recover if the DOM shifts (or the portal is down). That’s the gap we close in the next section (intent metadata + deterministic replay with targeted fallbacks).\n\n### **Step 2: Ask the agent to write down its intention so we can re-use it later**\n\nExploration gives us a working script, but it’s brittle because it only knows _what_ to click, not _why_. The fix was to capture **intents** to every action so the run can recover when the page shifts.\n\n![](https://www.skyvern.com/blog/content/images/2025/10/image---2025-10-17T132041.611.png)\n\nTo get to this intention, we generate 2 additional parameters at runtime: `user_detail_query` and `user_detail_answer` to capture the **essence** of the action (beyond the interaction itself)\n\n![](https://www.skyvern.com/blog/content/images/2025/10/image-3.png)\n\nThen, we pass it through another LLM call to reverse engineer the action into the following:\n\n![](https://www.skyvern.com/blog/content/images/2025/10/image-4.png)\n\nIf this fails at replay time, we don’t blindly guess a new selector—we reuse the **intention** to recover:\n\n1. Try an alternate selector for the same intention (looser match, nearby label, aria text)\n2. If the flow changed, ask the model **once**: “How do I ‘Select legal structure: Corporation’ on this page?”\n3. If we hit a dead end (downtime/error), fallback to the original prompt to decide what to do next\n\nAt this point, we have generated code that looks like this:\n\n![](https://www.skyvern.com/blog/content/images/2025/10/image-5.png)\n\n### **Step 3 - Run it on the cheap**\n\nNow that we have a plan and a fallback in place, subsequent runs are all using plain playwright — no LLM in the loop.\n\nWe benchmarked this across our customers and saw:\n\n* Average automation run time goes from 278.95s → 119.92s (2.3x faster)\n* Average run cost goes from $0.11 → $0.04 (2.7x cheaper)\n* And maybe more important than either: **runs are now deterministic.**\n\n![](https://www.skyvern.com/blog/content/images/2025/10/Screenshot-2025-10-16-at-14.18.17.png)\n\n![](https://www.skyvern.com/blog/content/images/2025/10/Screenshot-2025-10-16-at-14.18.07-1.png)\n\n# **How is this being used in the wild?**\n\nSkyvern’s “explore → replay” pattern is already running quietly inside a bunch of workflows that used to require brittle, human-maintained scripts. A few examples:\n\n1. **Invoice Downloading**\n\nAgents log into vendor or utility portals with 1000s of different accounts, navigate to the right billing period, and pull invoices. When layouts or date filters change, the intent metadata lets them recover automatically instead of failing.\n\n2. **Purchasing**\n\nTeams use it to automate repeat purchases — think renewing software licenses or buying supplies through the same vendor dashboard each month. The first run learns the checkout path, the fallbacks handle the variety of products, and the replays run deterministically, flagging if a price or SKU changes.\n\n3. **Data Extraction from Legacy Systems**\n\nSkyvern navigates authenticated dashboards, scrapes tables or PDFs, and pushes the structured output into into a database via Webhooks. If the DOM shifts, Skyvern reuses the same intention (“extract transaction rows”) to remap selectors.\n\n4. **Government Form Filling**\n\nFrom payroll registration to business license renewals, Skyvern handles long, multi-step government forms that occasionally break static scripts. Explore mode figures out the flow once replay mode repeats it safely.\n\n## **What’s next? Is it perfect today?**\n\nNot quite. The architecture works well, but there are still a few places where we can make it smarter and cheaper:\n\n1. **Analyze groups of runs when generating code**.\n\nRight now, we don’t aggregate insights across failures. Each replay fixes itself in isolation. If we could analyze _many_ runs together, spotting which selectors break, which flows diverge, we could automatically generalize better code and reduce the need for fallbacks. That’s especially useful for workflows that branch like trees (different inputs → different paths).\n\n2. **Cache data extractions**.\n\nDuring data extraction, we still rely on the LLM to “read” the page each time because many of our users want to both extract and summarize information at the same time. For example, if you ask Skyvern to pull the summaries of the top five posts on Hacker News, it currently parses the DOM from scratch. We’d like to trace _how_ the model found those elements (which selectors, which substrings) and reuse that mapping. That alone could make scraping and data-harvesting flows an order of magnitude cheaper.\n\n3. **Expose everything through the SDK.**\n\nWe think it will be valuable for developers using the Skyvern SDK to auto-generate these scripts for any ai actions / workflows they run, and use them automatically for subsequent runs. It currently requires a Skyvern server running, but soon it will be the default behaviour.\n\n## **Give it a try!**\n\nRun it via our [Skyvern Open Source](https://github.com/Skyvern-AI/Skyvern?ref=skyvern.com) or [Skyvern Cloud](http://app.skyvern.com/?ref=skyvern.com) versions and let us know what you think!", "url": "https://www.ycombinator.com/launches/Odo-skyvern-asking-ai-to-build-scrapers-should-be-easy-right", "votes": 12, "created_at": "2025-10-17T18:15:33.896Z" }, { "id": 90850, "title": "📊 Web Bench - A new way to compare AI Browser Agents", "tagline": "Expanding on the foundations of WebVoyager", "body": "**TL;DR:** [Web Bench](https://www.webbench.ai) is a new dataset to evaluate web browsing agents that consists of 5,750 tasks on 452 different websites, with [2,454 tasks being open sourced.](https://github.com/Halluminate/WebBench/tree/main) Anthropic Sonnet 3.7 CUA is the current SOTA, with the detailed [results here](https://eval.skyvern.com/?evaluation_target=skyvern_final_eval_skyvern_2_0&page=1).\n\nOver the past few months, Web Browsing agents such as [Skyvern](https://www.skyvern.com), [Browser-use](https://browser-use.com/) and [OpenAI's Operator (CUA)](https://operator.chatgpt.com/) have taken the world by storm. These agents have been used in production for a variety of tasks, from helping people apply to jobs, downloading invoices, and even doing SS4 filings for newly incorporated companies.\n\n![](https://blog.skyvern.com/content/images/2025/05/onlayer-gif.gif)\n\n_Skyvern attempting to purchase a product_\n\n![](https://blog.skyvern.com/content/images/2025/05/irs_demo.gif)\n\n_Skyvern attempting to fill out the IRS form_\n\nMost agents report state of the art performance, but we find that browser agents still struggle with a wide variety of tasks, particularly ones involving authentication, form filling and file downloading.\n\nThis is because the standard benchmark today ([WebVoyager](https://arxiv.org/abs/2401.13919)) focuses on read-heavy tasks and consists of **only** **643 tasks across only** **15 websites (out of 1.1 billion possible websites!).** While a great starting point, the benchmark does not capture the internet’s adversarial nature towards browser automation and the difficulty of tasks involving mutating of data on a website.\n\n![](https://blog.skyvern.com/content/images/2025/05/chase3-ezgif.com-video-to-gif-converter--1-.gif)\n\n_Can’t access [chase.com](http://chase.com)_\n\n![](https://blog.skyvern.com/content/images/2025/05/2025-05-08T02_12_10.137717_a_390786750312029848_recording-ezgif.com-video-to-gif-converter--1-.gif)\n\n_Can’t close a pop-up dialog_\n\nAs a result, we partnered with [Halluminate](https://www.halluminate.ai) and created a new benchmark to better quantify these failures. Our goal was to create a new consistent measurement system for AI Web Agents by expanding the foundations created by [WebVoyager](https://arxiv.org/abs/2401.13919) by:\n\n1. Expanding the number of websites from 15 → 452, and tasks from 642 -> 5,750 to test agent performance on a wider variety of websites\n2. Introduce the concept of READ vs WRITE tasks\n3. READ tasks involve navigating websites and fetching data\n * WRITE tasks involve entering data, downloading files, logging in, solving 2FA, etc, and were not well represented in the WebVoyager dataset\n4. Measure the impact of browser infrastructure (eg, access the websites, solve captchas, not crash, etc)\n\nWe’re excited to announce Web Bench, a new dataset to evaluate web browsing agents that consists of 5,750 tasks on 452 different websites, with [2,454 tasks being open sourced.](https://github.com/Halluminate/WebBench/tree/main)\n\n# Results TL;DR\n\n![](https://blog.skyvern.com/content/images/2025/05/Screenshot-2025-05-26-at-6.13.00---PM--1-.png)\n\n* All agents performed surprisingly poorly on write-heavy tasks (eg, logging in, filling out forms, downloading files), which implies that this is the area for the highest opportunity for growth\n* Skyvern 2.0 has the best performance for tasks in this category\n\n![](https://blog.skyvern.com/content/images/2025/05/Screenshot-2025-05-25-at-11.54.34---AM.png)\n\n* Agent performance for read-heavy tasks (eg, extracting information out of websites) was better than we expected\n* Anthropic’s CUA had the highest performance\n* The [entire run is **open source**](https://eval.skyvern.com/) and can be viewed to see the specific performance of each task\n\n# Dataset Creation\n\n![](https://blog.skyvern.com/content/images/2025/05/Screenshot-2025-05-25-at-11.49.12---AM.png)\n\nThe 452 websites are distributed across 17 primary categories. We sampled the benchmark websites from the [top 1000 websites in the world](https://dataforseo.com/free-seo-stats/top-1000-websites) measured by web traffic. We then cleaned this dataset of sites by removing:\n\n* repeat domains\n* sites without English translations\n* sites blocked by paywall\n\n# Results\n\n## Initial conditions\n\n* We ran OpenAI Operator with a human in the loop to establish a baseline for performance\n* We used consistent browser infrastructure (Skyvern’s infrastructure) when comparing the API-only models without a runtime to eliminate variables\n* We also ran Skyvern 2.0 on Browserbase to compare the impact of infrastructure, but found (surprisingly) that Skyvern’s infrastructure was able to reliably access more websites and encountered fewer anti-bot issues during navigation\n* Each agent was allowed a max of 50 steps per execution\n* Each result was validated by a human in the loop to assert evaluation data quality\n\n## Accuracy (Overall)\n\n![](https://blog.skyvern.com/content/images/2025/05/Screenshot-2025-05-25-at-11.53.34---AM--1-.png)\n\nThese results were a little bit surprising, so we decided to cut the data along 2 dimensions to understand where agents may falter:\n\n1. Read-only tasks (ie, extracting visible data from a particular website)\n2. Write heavy tasks (ie, logging into websites, filling out forms, downloading files)\n\n## Accuracy (Read-only tasks)\n\n![](https://blog.skyvern.com/content/images/2025/05/Screenshot-2025-05-25-at-11.54.34---AM--1-.png)\n\nRead-only tasks constitute tasks that involve agents going to different websites and navigating the sitemap until a particular answer/state has been found.\n\nUnsurprisingly, these results matched the WebVoyager dataset more closely, as the WebVoyager dataset was largely curated to help agents navigate websites and answer questions.\n\nThe biggest 2 sources of failures for read-heavy tasks are:\n\n* Navigation Issues (cannot figure out how to navigate a page, can’t solve popups)\n* Information extraction issues (doesn’t pull the correct information)\n\n## Accuracy (Write-heavy tasks)\n\n![](https://blog.skyvern.com/content/images/2025/05/Screenshot-2025-05-26-at-6.13.00---PM--2-.png)\n\nTasks involving:\n\n* filling out forms\n* logging in\n* solving 2FA\n* downloading files\n\nhad a much lower pass rate across the board.\n\nDigging a bit deeper into the failures, there were two culprits for the failures that popped up:\n\n* Incomplete execution (hallucinating that it’s achieved the goal when it has not)\n* Unable to identify the correct element to interact with (eg can’t close a popup dialog)\n\n![](https://blog.skyvern.com/content/images/2025/05/ezgif.com-speed-1.gif)\n\n_Can’t close subscription pop-up_\n\n![](https://blog.skyvern.com/content/images/2025/05/error2-ezgif.com-speed-1.gif)\n\n_Unable to find and click coupon buttons_\n\nThese issues manifest as agents making adverse changes when filling out forms, or optimistically assuming that clicking a “Submit” button completes the task, when in reality, a captcha appeared that now needs to be solved.\n\nThis issue is very similar to the phenomenon observed in coding agents, where smarter models try to “overhelp” with code changes by either making changes to unrelated parts of the codebase or repeatedly suggesting incorrect things because they’re missing some important context.\n\n## Agent Failure Modes - Deepdive\n\n![](https://blog.skyvern.com/content/images/2025/05/Screenshot-2025-05-25-at-11.56.30---AM.png)\n\nDigging deeper, we were able to qualify the overall failures into two buckets:\n\n1. Agent Failures (eg, Agents hallucinated/made poor decisions or didn’t interact with important elements)\n2. Infrastructure failures (eg, Agent can’t access the website, solve a captcha to log in)\n\n### Agent issues\n\nThe 4 biggest categories of agent errors are:\n\n* Navigation Issue (cannot figure out how to navigate a page, can’t solve popups)\n* Incomplete execution (hallucinating that it’s achieved the goal when it has not)\n* Time-outs (exceeds step limits)\n* Information extraction issue (doesn’t pull the correct information)\n\n### **Infrastructure issues**\n\nThe 3 biggest categories of infrastructure issues are:\n\n* Proxy (Failed to access website/website blocked)\n* Captcha (Verification required to proceed and the infrastructure is unable to solve it)\n* Login/Authentication (Google Auth detecting you’re a bot)\n\nThese findings imply that the browser infrastructure powering the agents is equally as important as the quality of the agent itself.\n\n# Other interesting characteristics\n\nWhile accuracy is the most important characteristic of a web browsing agent, there is also a desire to get “faster” and “cheaper” agents.\n\nFast and cheap agents can be characterized by tracking the following metrics:\n\n1. Runtime duration\n2. Number of steps\n\nWhile the right pricing model for browser agents hasn’t emerged yet, this data gives an important insight into whether pricing per hour (common amongst hosted browsers + older robotic process automation) and pricing per step (common amongst computer use APIs) is the right methodology.\n\n## Agent Runtime Duration\n\n![](https://blog.skyvern.com/content/images/2025/05/Screenshot-2025-05-28-at-10.14.41---AM.png)\n\nThis metric is important for a few latency-sensitive market segments/situations:\n\n1. Copilot-like products where a human is supervising 1 or many agents executing in parallel\n2. Phone agents referencing information / doing real-time lookups while a user is talking\n3. Websites aggregating information in real-time to show to the user(e.g., looking up flight or domain name availability)\n\n## Steps\n\n![](https://blog.skyvern.com/content/images/2025/05/Screenshot-2025-05-25-at-10.47.56---PM.png)\n\nMost web browsing agents’ costs scale with the number of steps (i.e., page scans) required to complete a specific task.\n\nAgents may use a varied number of steps for a few reasons:\n\n1. Most agents use different architectures to batch actions together to minimize the number of steps\n2. Agents eager to solve the problem may use a lot of steps in error situations to try to independently resolve their issues (e.g., chatting with support for solutions instead of terminating early)\n3. Agents using pessimistic approaches to reduce hallucinations may invalidate batches of actions whenever the website changes after a particular action (i.e., filling in an address field might invalidate the action plan for the rest of the page)\n\n# Next Steps\n\n* We intend to benchmark [Claude 4](https://www.anthropic.com/news/claude-4), [Operator O3](https://openai.com/index/o3-o4-mini-system-card-addendum-operator-o3/), [UI-TARs](https://github.com/bytedance/UI-TARS), and [Mariner API](https://deepmind.google/models/project-mariner/)\n* We intend to expand the benchmark to cover more categories of websites beyond the top 1000 websites in the world\n* We intend to expand this dataset to include other languages to assert multi-lingual browser agent performance\n\n# FAQ\n\n### Why not include a bigger cohort of websites?\n\nThe goal of the first version of Web Bench was to start with a reasonably large number of websites + tasks.\n\nThe next few versions will expand: (1) the number of websites encompassed by this benchmark, (2) the language of websites, and (3) the categories of websites\n\n### Why were only a subset of vendors benchmarked?\n\nThe overall cost to run this benchmark was approximately $3,000 per agent for human evaluations. This cost is high enough to be prohibitive of testing every single agent on the market.\n\nThe halluminate team plans to release an open-source automated evaluation harness to allow teams to self-serve testing against [Web Bench](https://www.webbench.ai). We leave this for future work.\n\nIf you’d like to submit benchmark results for your own agent, please [contact the Halluminate team](mailto:jerry@halluminate.ai)\n\n### Where can I read the technical deepdive?\n\nFor a comprehensive technical debrief on the dataset creation and evaluation methodology, check out the Halluminate team writeup [here](https://halluminate.ai/blog/benchmark).\n\n### Did anything funny happen along the way?\n\nYes.\n\nWe had examples of Skyvern chatting with GitHub’s AI Support bot\n\n![](https://blog.skyvern.com/content/images/2025/05/image--67-.png)\n\nBrowser Use googled how to evade Cloudflare captchas\n\n![](https://blog.skyvern.com/content/images/2025/05/image--68-.png)\n\n", "url": "https://www.ycombinator.com/launches/NdK-web-bench-a-new-way-to-compare-ai-browser-agents", "votes": 18, "created_at": "2025-05-29T18:26:39.628Z" }, { "id": 89234, "title": "🐉 Skyvern MCP Server: let agents control your browser", "tagline": "Meet Skyvern, an open source AI agent that can control your browser and complete complex tasks", "body": "We were playing around with MCPs over the weekend and thought it would be cool to build an MCP that lets Claude / Cursor / Windsurf control your browser: \n\nJust for context, we’re building Skyvern, an open-source AI Agent that can control and interact with browsers using prompts, similar to OpenAI’s Operator.\n\nThe MCP Server can:\n\n* This allows Claude to navigate to docs websites/stack overflow and look up information like the top posts on hackernews\n\n![uploaded image](/media/?type=post&id=89234&key=user_uploads/1120458/9a0259b8-ebb7-44c4-9f4d-a881c31c8d10)\n\n* This allows Cursor to apply for jobs, fill out contact forms, log in and download files, etc.\n\n ![uploaded image](/media/?type=post&id=89234&key=user_uploads/1120458/f9c936a8-88de-47cc-8209-7983754fcb8a)\n\n* Connect Windsurf to take over your Chrome while running Skyvern in “local” mode\n\n![uploaded image](/media/?type=post&id=89234&key=user_uploads/1120458/40ca033c-fa32-4c7f-bda0-7cd072ae9212)\n\nWe built this mostly for fun, but we can see this being integrated into AI agents to give them custom access to browsers and execute complex tasks like booking appointments, downloading your electricity statements, looking up freight shipment information, etc.", "url": "https://www.ycombinator.com/launches/NDG-skyvern-mcp-server-let-agents-control-your-browser", "votes": 8, "created_at": "2025-04-03T18:28:44.864Z" }, { "id": 86895, "title": "Skyvern 2.0 - State-of-the-art web navigation with 85.8% on WebVoyager Eval", "tagline": "Open source AI Web Agent", "body": "We’ve been working hard, cooking up something new to share with you all!\n\nSkyvern 2.0 **scored state-of-the-art 85.85%** on the WebVoyager Eval.\n\nThis is the best-in-class performance of all WebAgents, giving advanced closed-source web agents like Google Mariner a run for their money.\n\n![uploaded image](/media/?type=post&id=86895&key=user_uploads/1120458/4f04949f-e01d-42cb-b568-a2460f7ac7f5)\n\n### TL;DR\n\n* **Real-World Tests:** We ran all of the tests in Skyvern Cloud to get a better representation of autonomous browser operations (ie, they didn’t run on any local machines)\n* **Open-Sourced Results:** All of the runs can be [seen here](https://eval.skyvern.com/) through our UI.\n* **We’re just getting started**. Try [Skyvern Cloud](https://app.skyvern.com/) or [Skyvern Open Source](https://github.com/Skyvern-AI/Skyvern) out for yourself and see Skyvern in action!\n\n## Agent Architecture\n\nAchieving this SOTA result required expanding Skyvern’s original architecture. Skyvern 1.0 involved a single prompt operating in a loop both making decisions and taking actions on a website. This approach was a good starting point but scored \\~45% on the WebVoyager benchmark because it had insufficient memory of previous actions and could not do complex reasoning.\n\nTo solve this problem, we created a self-reflection feedback loop within Skyvern. This resulted in 2 main changes:\n\n1. We added a “Planner” phase, which could decompose very complex objectives down into smaller achievable goals\n2. This allowed Skyvern to have a working memory of things it had completed and things that were still waiting to be finished\n * This allows Skyvern to work with long, complex prompts without increasing the hallucination rate\n3. We added a “Validator” phase, which confirmed whether or not the original goals the “Planner” generates are successfully completed\n4. This acts as a supervisor function to confirm that the Task executor is achieving its objectives as expected and report any errors/tweaks back to the Planner so it can make adjustments in real-time as needed\n\n![](https://blog.skyvern.com/content/images/2025/01/architecture.png)\n\n## Test Setup\n\nAll tests were run in [Skyvern Cloud](https://app.skyvern.com/) with an **async cloud browser** and used a combination of GPT-4o and GPT-4o-mini as the primary decision-making LLMs. The goal of this test is to assert real-world quality — the quality represented by this benchmark is the same as what you would experience with [Skyvern](https://app.skyvern.com/)’s browsers running asynchronously.\n\n💡 **Why is this important?** Most benchmarks are run on local browsers with a relatively safe IP address and an impressive browser fingerprint. This is not representative of how Autonomous agents will run in the cloud, and we wanted our benchmark to represent how agents would behave in production\n\nIn addition to the above, we’ve made a few minor tweaks to the dataset to bring it up to date:\n\n1. We’ve removed 8 tasks from the dataset because the results are no longer valid. For example, one of the tasks asked to go to [apple.com](http://apple.com) and check when the Apple Vision Pro will be released — in 2025, it’s already been released and forgotten\n2. Many of the flight/hotel booking tasks referenced old dates. We updated both the prompt and the answer to more modern dates for this evaluation\n\n🔍 **For the curious**:\n\nThe full dataset can be seen here: \n\nThe full list of modifications can be seen here: \n\n## Test Results\n\nWe’re doing something out of the ordinary. In addition to the results, we’re making our entire benchmark run public.\n\n💡 **Why is this important?** Most benchmarks are run behind closed doors, with impressive results being published without any accompanying material to verify the results. This makes it hard to understand how things like hallucinations or website drift over time play an effect on agent performance\n\nWe believe this isn’t aligned with our open source mission, and have decided to publish the entire eval results to the public.\n\n📊 All individual run results can be seen here: \n\n🔍 The entire Eval dataset can be seen here: \n\n## Limitations of the WebVoyager benchmark\n\nThe WebVoyager benchmark is a comprehensive benchmark testing a variety of prompts on 15 different websites. While this acts as a good first step in testing Web agents, this only captures 15 hand-picked websites of the millions of active websites on the internet.\n\nWe think there is tremendous opportunity here to better evaluate web agents against one another with a more comprehensive benchmark similar to SWE-Bench.\n\n## What’s on the horizon\n\nBrowser automation is still a nascent space with tons of room for improvement. While we’ve achieved a major milestone in agent performance, a few important issues are next to be solved:\n\n1. Can we improve Skyvern’s reasoning to operate efficiently in situations with more uncertainty? Examples include vague prompts, ambiguous or highly complex websites/tools, websites with extremely poor UX (legacy portals)\n2. Can we give Skyvern access to more tools so it can effectively log into websites, make purchases, and behave more like a human?\n3. Can we have Skyvern better memorize things it has already done in the past so it can do them again at a lower price point?\n\n## References\n\n* [Google Mariner Report](https://deepmind.google/technologies/project-mariner/)\n* [Claude Computer Use Report](https://docs.anthropic.com/en/docs/build-with-claude/computer-use)\n* [WebVoyager Report](https://arxiv.org/html/2401.13919v4)\n* [WILBUR Report](https://arxiv.org/html/2404.05902v1)\n* [AgentE Report](https://arxiv.org/html/2407.13032v1)\n* [HCompany Report](https://www.hcompany.ai/blog/a-research-update)", "url": "https://www.ycombinator.com/launches/MbX-skyvern-2-0-state-of-the-art-web-navigation-with-85-8-on-webvoyager-eval", "votes": 10, "created_at": "2025-01-16T16:48:00.855Z" }, { "id": 86198, "title": "🐉 Skyvern's Contact Form Agent: Automate contact form submissions with AI", "tagline": "Cold outbound is the worst. Try this new form of outbound instead!", "body": "Companies use [Skyvern](https://www.skyvern.com/) to automate browser-based workflows, doing tasks like form-filling, invoice downloading, and more. One emerging use case for Skyvern has been navigating to thousands of different websites and filling out contact forms... to scale a new untapped outbound channel.\n\nAnd this strategy blew our minds. Why:\n\n1. Companies with contact forms often, guess what, want to be contacted\n2. Contact form submissions often go to a specific team dedicated to help with inquiries\n3. You avoid all of the craziness that comes with setting up email infrastructure: buying up domains, warming inboxes, worrying about bounce rates and other email nuisances\n4. You can send as many as you want whenever you want—forget limiting the number you send per day to increase deliverability\n\nThe demand and success of this use case led us to create a dedicated agent tailored to it: Skyvern Contact Forms. And this is why I’m here, to share this sick outbound strategy and let you know we can help :)\n\nTry it out at [**https://app.skyvern.com/forms**](https://app.skyvern.com/forms), let us know what you think!\n\nLike what you see and want bulk pricing? [**I’d love to chat.**](https://meetings.hubspot.com/skyvern/demo)\n\n![uploaded image](/media/?type=post&id=86198&key=user_uploads/1120458/2e37309e-035e-41cd-b89a-c8552a5019d6)\n\n", "url": "https://www.ycombinator.com/launches/MQI-skyvern-s-contact-form-agent-automate-contact-form-submissions-with-ai", "votes": 8, "created_at": "2024-12-10T18:35:41.619Z" }, { "id": 81634, "title": "🐉 Skyvern Cloud - Open source AI Agent to automate browser based workflows", "tagline": "The easiest way to automate similar tasks on a lot of different websites", "body": "### TL;DR\n\n[Skyvern](https://skyvern.com/) helps companies automate browser-based workflows using AI. We provide a simple API endpoint to fully automate manual workflows, replacing brittle or unreliable scripts. We’re open source — check out [our repository here](https://github.com/Skyvern-AI/Skyvern). **We just launched Skyvern Cloud — check it out at **[**app.skyvern.com**](https://app.skyvern.com)**.**\n\n**—**\n\n### 🧙‍♂️🐉 Skip the boring explanation, show me the magic!\n\n**We have good news!** We just launched Skyvern Cloud. You can go to [app.skyvern.com](http://app.skyvern.com) and try playing around with it! You’ll get $5 of credit when signing up and setting up a payment method\n\nHere’s an example of it filling out the job application form and applying to Lever’s job demo website:\n\n![uploaded image](/media/?type=post&id=81634&key=user_uploads/1120458/b3e666b8-5e36-4e7e-91ac-1120de55cd39)\n\n### 🤔 What are browser workflows?\n\nMost companies we’ve talked to have manual or semi-automated workflows that support their core product. Many of these workflows started manually as companies were getting off the ground, [doing things that don’t scale](https://paulgraham.com/ds.html), and evolved to occupy a larger amount of human capital or automated with unreliable scripts.\n\n**Example 1 — Downloading invoices on a large number of websites**\n\nLet’s say it’s the end of the month, and your accountant is hounding you for invoices for all your transactions this month. Traditionally, you would have to log into each portal and download it yourself. Skyvern can help automate this for you! **PS: If you’d like to automate a complex flow like this, we’d [love to help get you set up.](https://meetings.hubspot.com/suchintan)**\n\n![uploaded image](/media/?type=post&id=81634&key=user_uploads/1120458/49e1d569-5a46-4728-a3ec-6a0a4d8c40aa)\n\n**Example 2 — Filling out forms + uploading documents on Government websites**\n\nLet’s say you’re an accounting company or a law firm, and you want to help your clients get set up for Payroll in California — you wouldn’t want to go get registered for EDD manually for each potential customer.. and in each state! Call in a favor from Skyvern to help automate this.\n\n![uploaded image](/media/?type=post&id=81634&key=user_uploads/1120458/66682edb-bdc4-4a03-b157-235501356bfc)\n\n**Example 3 — Automating materials procurement from commerce websites**\n\nLet’s say you’re a B2B marketplace for car parts, and your users submit orders from multiple vendors with you. A subset of vendors you work with don’t support ordering via an API, so you have a human in the loop to transact orders for those vendors. Here’s an example of Skyvern navigating to [finditparts.com](http://finditparts.com/) and ordering an airbag — all via an API call!\n\n![uploaded image](/media/?type=post&id=81634&key=user_uploads/1120458/f4630123-b7d1-4fff-aedb-31dd3ea11f05)\n\n**Example 4 —** **Completing dynamic multi-step workflows**\n\nLet’s say that you’re trying to automate complex workflows, such as generating insurance quotes. Skyvern is able to navigate each step until its specified goal is achieved — even if the steps change depending on the user’s situation. Here’s an example of Skyvern navigating to [Geico.com](http://geico.com/) and generating an auto insurance quote with your personalized information.\n\n![uploaded image](/media/?type=post&id=81634&key=user_uploads/1120458/f800fef8-1a6d-4345-ba5a-043a154ad488)\n\n### 🤩 How does it work??\n\nSkyvern uses LLMs to become an AI Agent capable of interacting with websites like you or I would — all via a simple API call.\n\nTraditional approaches required writing custom scripts for websites, often relying on [DOM parsing](https://developer.mozilla.org/en-US/docs/Web/API/Document_Object_Model) and [XPath-](https://en.wikipedia.org/wiki/XPath)based interactions, which would break whenever the website layouts changed.\n\nSkyvern’s prompt → automation breaks these constraints and relies on multi-modal LLMs to parse items in the viewport and interact with them the way a human would.\n\nThis approach gives us a few advantages:\n\n1. Skyvern can take a single prompt (i.e., “Download an invoice from the order history page”) and repeat it across a large number of similar websites. This would traditionally require one script per website — making tackling the long tail of website interactions very challenging\n2. Skyvern is able to operate on websites it’s never seen before, as it’s able to map visual elements to actions necessary to complete a workflow without any customized code\n3. Skyvern is resistant to website layout changes, as there are no pre-determined XPaths or other selectors our system is looking for while trying to navigate\n4. We’re able to circumvent or navigate through many bot detection methods, as many of them rely on looking for outlier behavior\n5. We rely on LLMs to reason through interactions to make sure we can cover complex situations. Examples include:\n * If you wanted to get an auto insurance quote from Geico, the answer to a common question, “Were you eligible to drive at 18?” could be inferred from the driver receiving their license at age 16\n * If you were doing competitor analysis, it’s understanding that an Arnold Palmer 22 oz can at 7/11 is almost definitely the same product as a 23 oz can at Gopuff (even though the sizes are slightly different, which could be a rounding error!)\n\n### 😮 And you’re open source?\n\nYes! You can [check out our code here](https://github.com/Skyvern-AI/Skyvern). We’re open source for two big reasons:\n\n1. It allows developers to be able to look at, understand, and dive deep into Skyvern’s implementation details to (1) expand their capabilities by adding support for new functionality and (2) decode why they’re doing what they’re doing.\n2. It allows security-minded enterprises to escape “security theater” and keep data on-prem by self-hosting Skyvern.\n 1. If this is at all interesting to you, [let’s chat](https://meetings.hubspot.com/suchintan?uuid=9cb375c0-cb2c-4900-b22d-f52baa74dfb2).\n\n### Our Ask\n\nDo you have any complex workflows that you’d love to automate? We’d love to chat! Shoot me an email at [suchintan@skyvern.com](mailto:suchintan@skyvern.com) or grab some time via [my calendar.](https://meetings.hubspot.com/suchintan)", "url": "https://www.ycombinator.com/launches/LEg-skyvern-cloud-open-source-ai-agent-to-automate-browser-based-workflows", "votes": 35, "created_at": "2024-06-20T14:46:54.762Z" }, { "id": 77665, "title": "🐉 Skyvern - Automate Browser-based workflows with AI", "tagline": "Automate manual workflows within your company with an AI Agent", "body": "### TL;DR\n\nSkyvern helps companies automate browser-based workflows using LLMs and computer vision. We provide a simple API endpoint to fully automate manual workflows, replacing brittle or unreliable scripts.\n\n### 🤔 What are browser workflows?\n\nMost companies we’ve talked to have manual or semi-automated workflows that support their core product. Many of these workflows started manually as companies were getting off the ground, [doing things that don’t scale](https://paulgraham.com/ds.html), and evolved to occupy a larger amount of human capital or automated with unreliable scripts.\n\n**Example 1 — Automating materials procurement from commerce websites**\n\nLet’s say you’re a B2B marketplace for car parts, and your users submit orders from multiple vendors with you. A subset of vendors you work with don’t support ordering via an API, so you have a human in the loop to transact orders for those vendors. Here’s an example of Skyvern navigating to [finditparts.com](http://finditparts.com) and ordering an airbag — all via an API call!\n\n![uploaded image](/media/?type=post&id=77665&key=user_uploads/1120458/055a1d70-8ed2-428d-8104-740f7632d3e7)\n\n**Example 2 —** **Completing complex multi-step workflows**\n\nLet’s say you’re trying to automate complex workflows such as generating insurance quotes. Skyvern can navigate each step until its specified goal is achieved — even if the steps change depending on the user’s situation. Here’s an example of Skyvern navigating to [Geico.com](http://Geico.com) and generating an auto insurance quote with your personalized information.\n\n![uploaded image](/media/?type=post&id=77665&key=user_uploads/1120458/747160ae-eb91-48c0-977d-830074a5b93b)\n\n### 🤩 How does it work??\n\nOur focus is bringing stability to browser-based workflows. We leverage LLMs to create an AI Agent capable of interacting with websites like you or I would — all via a simple API call.\n\nTraditional approaches required writing custom scripts for websites, often relying on [DOM parsing](https://developer.mozilla.org/en-US/docs/Web/API/Document_Object_Model) and [XPath-](https://en.wikipedia.org/wiki/XPath)based interactions which would break whenever the website layouts changed.\n\nSkyvern operates like a human — increasing reliability by not relying on fragile scripts, instead relying on computer vision to parse items in the viewport and interact with them the way a human would.\n\nThis approach gives us a few advantages:\n\n1. Skyvern can operate on websites it’s never seen before, as it’s able to map visual elements to actions necessary to complete a workflow, without any customized code\n2. Skyvern is resistant to website layout changes, as there are no pre-determined XPaths or other selectors our system is looking for while trying to navigate\n3. We’re able to circumvent or navigate through many bot detection methods as many of them rely on allowing people to access the websites\n4. We rely on LLMs to reason through interactions to ensure we can cover complex situations. Examples include:\n * If you wanted to get an auto insurance quote from Geico, the answer to a common question “Were you eligible to drive at 18?” could be inferred from the driver receiving their license at age 16\n * If you were doing competitor analysis, it’s understanding that an Arnold Palmer 22 oz can at 7/11 is almost definitely the same product as a 23 oz can at Gopuff (even though the sizes are slightly different, which could be a rounding error!)\n\n### Our Ask\n\nDo you have any complex workflows that you’d love to automate? Were you [doing things that don’t scale](https://paulgraham.com/ds.html) that you now need to scale? We’d love to chat! Shoot me an email at [suchintan@skyvern.com](mailto:suchintan@skyvern.com) or grab some time via [my calendar.](https://meetings.hubspot.com/suchintan)", "url": "https://www.ycombinator.com/launches/KCf-skyvern-automate-browser-based-workflows-with-ai", "votes": 69, "created_at": "2024-01-19T17:34:36.013Z" }, { "id": 72744, "title": "🐉 Wyvern AI - Increase revenue in marketplaces with better product ranking", "tagline": "With Wyvern, not only do marketplaces get better product ranking, they get the ability to fine-tune the ranking with unique insights about their business", "body": "**TL;DR:** [Wyvern.ai](https://www.wyvern.ai?utm_source=yc&utm_medium=yc_launch_email&utm_campaign=launch_yc&utm_id=10) is a machine learning platform for marketplaces to help them optimize their product ranking. We do this by providing an API that’s integrated with their product catalog, that optimizes product ranking based on factors such as popularity, relevance, and personalization. In addition, we also provide a framework where customers can fine-tune their product ranking models with custom insights and tailor them to their business. We built this exact platform before, and it unlocked over $100M of value.\n\n## 💸 Product Ranking: **Unleashing Growth Potential**\n\nMarketplaces, such as Amazon, eBay, or Etsy, often have millions of products listed by various sellers, making it crucial to determine how to present these products to users effectively. These companies have teams of data scientists dedicated to improving the quality of product ranking, factoring in things like personalization (ie individual user preferences), relevance, and general product popularity.\n\nOur experience building Faire and Gopuff’s machine learning platform highlighted the value that product ranking derives for the business. Through successive iterations of machine learning models, we witnessed impacts exceeding 10% (equivalent to over $100 million) for these businesses.\n\nA notable portion of these impactful iterations involved providing machine learning models with access to unique insights into user behaviour. For instance, at Faire, gaining a better understanding of what products their retailers stocked in their stores enabled tailoring the shopping experience to suit each store's distinct characteristics.\n\nOther larger marketplaces had their own insights. At eBay, incorporating buyer-seller shipping logistics information into their product ranking models produced notable improvements. Similarly, at Etsy, the users' initial interactions with the site indicate their shopping intention, allowing their product ranking models to transform the entire homepage to deliver a personalized experience.\n\n## 💣 Challenges to navigate along the way\n\nWhile larger marketplaces have the leverage to iterate on their product ranking, smaller marketplaces tend to buy off-the-shelf solutions, allowing them to bootstrap and search for product-market-fit.\n\nOnce a marketplace establishes product-market-fit, transitioning from an off-the-shelf product ranking solution to a highly customizable one is deceptively challenging and requires substantial engineering investment:\n\n1. **Latency:** Effective product ranking models perform better when they can evaluate a large number of products. However, as the number of products increases, so does the user-facing latency. Maintaining fast browsing experiences while running machine learning models on hundreds of products in under 100 milliseconds is a demanding technical hurdle.\n2. **Customizability:** The value derived from machine learning models in product ranking heavily relies on incorporating unique user behavior within the marketplace. Therefore, the system needs to be highly flexible, enabling data science teams to explore and implement new insights they acquire about users.\n3. **Observability:** The importance of data quality and accuracy cannot be overstated in machine learning. Ensuring that that the data used for model training and evaluation is reliable, enabling effective decision-making and accurate product ranking.\n\n## 🐉 Solution: [Wyvern AI](https://www.wyvern.ai?utm_source=yc&utm_medium=yc_launch_email&utm_campaign=launch_yc&utm_id=10)\n\nWyvern’s machine learning platform solves these problems. Not only do we give marketplaces a model that immediately improves their product ranking, we also provide:\n\n1. The ability to iterate on and fine-tune the model by expanding its inputs, such as relevance, personalization, popularity, and more.\n2. The ability to adjust the model's objective, allowing it to optimize conversion, GMV, revenue, profit, or a combination thereof, based on the current focus of the business.\n3. An observability layer is provided to marketplaces, offering essential capabilities for debugging and evaluate historic predictions. This layer also facilitates transforming the data into a feedback loop, enabling continuous improvement of the performance of real-time machine learning models.\n\nThis allows our customers to dedicate data scientists to iterate on models within their marketplace, allowing them to continually improve the quality of product ranking across all of their surfaces.\n\n![uploaded image](/media/?type=post&id=72744&key=user_uploads/1120458/639a4715-c00e-430a-94c3-f5b122dfe255)\n\n## 🙏 Our asks\n\n* Are you a marketplace company? We would love to chat.\n* Do you know anyone working at growing marketplaces? We’d love an introduction.\n* If you have any questions, you can reach us at [suchintan@wyvern.ai](https://mailto:suchintan@wyvern.ai)", "url": "https://www.ycombinator.com/launches/IvI-wyvern-ai-increase-revenue-in-marketplaces-with-better-product-ranking", "votes": 18, "created_at": "2023-07-05T13:51:14.223Z" } ] }, { "id": 28996, "name": "Sola", "slug": "sola", "former_names": [ "AdMagic" ], "small_logo_thumb_url": "https://bookface-images.s3.amazonaws.com/small_logos/0ab41e90328883754ecff618e23fa11d9632ee43.png", "website": "https://www.sola-solutions.com/", "all_locations": "New York City, NY, USA", "long_description": "Sola is copilot for workflow automation. \r\n\r\nWe help automation-minded teams build robust robotic agents to handle tedious tasks via LLMs and computer vision, smoothly integrated into their standard workflow. \r\n\r\nOur tooling makes it easy for companies to automate data entry/scraping in manual filing and processing flows. Go from a screen recording to a bot in any desktop or browser workflow; develop logic and watch logic self-develop in the face of new events!", "one_liner": "Copilot for Robotic Process Automation", "team_size": 5, "industry": "B2B", "subindustry": "B2B", "launched_at": 1690480911, "tags": [ "Robotic Process Automation", "Workflow Automation", "Automation", "AI" ], "tags_highlighted": [], "top_company": false, "isHiring": true, "nonprofit": false, "batch": "Summer 2023", "status": "Active", "industries": [ "B2B" ], "regions": [ "United States of America", "America / Canada" ], "stage": "Growth", "app_video_public": false, "demo_day_video_public": false, "app_answers": null, "question_answers": null, "url": "https://www.ycombinator.com/companies/sola", "api": "https://devasheeshg.github.io/yc-api/batches/summer-2023/sola.json", "year_founded": 2023, "city": "New York City", "country": "US", "linkedin_url": "https://www.linkedin.com/company/98128293/", "x_url": "https://twitter.com/Sola_Solutions", "fb_url": null, "cb_url": null, "github_url": null, "logo_url": "https://bookface-images.s3.amazonaws.com/small_logos/0ab41e90328883754ecff618e23fa11d9632ee43.png", "app_video_url": null, "dday_video_url": null, "company_photos": [], "primary_partner": { "name": "Diana Hu", "url": "https://www.ycombinator.com/people/diana-hu" }, "founders": [ { "user_id": 1317145, "full_name": "Jessica Wu", "title": "Founder", "founder_bio": "", "is_active": true, "linkedin_url": "https://linkedin.com/in/jessica-wu-9824431ab", "x_url": "https://twitter.com/jess123555", "avatar_thumb_url": "https://bookface-images.s3.us-west-2.amazonaws.com/avatars/48d7a2d5cdcfcaa80fe92e70128a64d150ff6f6d.jpg" }, { "user_id": 787351, "full_name": "Neil Deshmukh", "title": "Founder", "founder_bio": "", "is_active": true, "linkedin_url": "https://linkedin.com/in/neildeshmukh", "x_url": "https://twitter.com/NeilDeshmukh", "avatar_thumb_url": "https://bookface-images.s3.us-west-2.amazonaws.com/avatars/da3fc432de1d9b29f157266ba4b97c8cd6fa5738.jpg" } ], "jobs": [], "news": [ { "title": "Sola Solutions raises $21 million for enterprise AI | Fortune", "url": "https://fortune.com/2025/08/14/sola-solutions-ai-startup-raised-21-million-andreessen-sarah-guo/", "date": "Aug 14, 2025" } ], "launches": [ { "id": 73648, "title": "🤖 Sola - Build AI agents in copilot", "tagline": "Meet Sola, your co-pilot for Robotic Process Automation!", "body": "**tl;dr:** _Sola is copilot for Robotic Process Automation. We help automation-minded teams visually script bots that emulate human behavior to automate data entry, scraping, and filing flows using LLMs and computer vision._\n\nHey everyone! We’re Jess and Neil, the founders of Sola.\n\n![uploaded image](/media/?type=post&id=73648&key=user_uploads/1317145/8edcf728-fa9a-4c2d-be38-a425e3a3a809)\n\nConsider a scenario: A manual, repetitive workflow demands that you extract information from a customer ticket, use this to lodge a legal application within their respective state, and then retain and store the resulting confirmation details.\n\nDeveloping engineering solutions for such workflow automation often brings unexpected overhead:\n\n* Engineers must map all error paths before deployment, often necessitating weeks or even months of rigorous testing, with frequent reengineering required as elements evolve — _what happens when the government filing sites crash or change?_\n* Continued investment in monitoring is crucial to maintain a working solution.\n* Increasingly tight security measures in browser and desktop applications make request-based automation increasingly challenging.\n\nConsidering these challenges in our scenario of building out the state-specific application processes, it makes more sense to automate from a _client-side perspective_.\n\nHowever, existing process automation solutions are reliant on integrations, non-generalizable, not self-healing, and often tough to use…\n\n### The Solution 🦾\n\nNow, **imagine going from a screen recording to a process-automating bot in seconds!** Sola RPA can do exactly this, empowering a future where data entry, scraping, and filing tasks are no longer a hassle, but handled efficiently by agents and citizen developers. Structured as a co-pilot tool, Sola bots are seamlessly integrated into the existing workflow, learning new branches and behavior automatically!\n\nWith the above scenario, internal tooling teams are able to now immediately deploy bots for each state-specific filing workflow, where the bot learns automatically from the human stepping in as it runs into errors.\n\nUsing state-of-the-art AI and computer vision, our tooling makes it easy for companies to automate data entry/scraping in legal, financial, and healthcare flows. We’ve built our tool with intuitive usability at its core, allowing users to **visually script** bots in a Figma-esque workflow that can extrapolate to new data (think Excel macros, but generalized)!\n\n![uploaded image](/media/?type=post&id=73648&key=user_uploads/1317145/64327f3b-dab9-4b3f-b7fb-96a9e44e4fcc)\n\nCheck out some of our case studies! For example,\n\n* We help an SMB unicorn [deploy bots for legal incorporation flows](https://www.sola-solutions.com/product/legal-filing) in each state.\n* We help a hedge fund [automate account interactions in desktop financial software](https://www.sola-solutions.com/product/desktop), where there’s no API to interface with, and [schedule scraping of unstructured data](https://www.sola-solutions.com/product/knowledge-extraction) to track expirations of securities.\n* We work with a scaling neobank to [scrape information on legacy financial government portals](https://www.sola-solutions.com/product/scraping-systems) for loan assessments from databases.\n\n![uploaded image](/media/?type=post&id=73648&key=user_uploads/1317145/b3692f9c-19bc-4256-87fd-f7cc0c4b9cab)\n\n### The Team\n\nJess and Neil have been best friends since their freshman year at MIT, and have worked together a ton even before Sola.\n\nNeil has researched CV and LLMs while studying CS at MIT (at the MIT-IBM Watson + Quest for Intelligence Labs) focusing on multimodal RL models, and built out ML at startups in areas including computational biology, fraud detection, healthcare, and agriculture.\n\nJess has studied math and CS at MIT, most recently having spent her time working in finance (at Citadel Securities, Goldentree AM, Thrive Capital, and some other hedge funds), and she’s also worked on startups in med-tech, fashion, and crypto, doing a mix of design work and engineering.\n\nDrawing from their wealth of experience in AI research, engineering, and design, as well as first-hand frustration with repetitive workflows in finance and healthcare, they’re on a mission to tackle the challenge of solving tedious flows in an intuitive way!\n\n### The Ask\n\nWe’d love to help automate your workflows!\n\nIf you’re a company thinking about robotic process automation or manual outsourcing as you scale, reach out to [founders@sola-solutions.com](mailto:founders@sola-solutions.com) and we’d love to chat about working together, as we onboard a limited number of design partners early on! :)", "url": "https://www.ycombinator.com/launches/J9s-sola-build-ai-agents-in-copilot", "votes": 691, "created_at": "2023-08-01T15:57:28.435Z" } ] }, { "id": 29137, "name": "Opensteer", "slug": "opensteer", "former_names": [ "Thumbnail Game", "TrendUp", "trendup", "Curatle", "dinogent", "Curatle", "Oversteer" ], "small_logo_thumb_url": "https://bookface-images.s3.amazonaws.com/small_logos/38b0ef5314d328b72aff9a8d9829350f181b5ace.png", "website": "https://opensteer.com", "all_locations": "Los Angeles, CA, USA; Remote", "long_description": "Opensteer creates specialized AI agents in cloud computers that can integrate with anything. Each agent gets customizable browser access, tools, instructions, and guardrails, so it can work 24/7 exactly the way you need.", "one_liner": "Specialized agents for real work", "team_size": 1, "industry": "B2B", "subindustry": "B2B", "launched_at": 1689032123, "tags": [ "Robotic Process Automation", "B2B", "API", "Automation", "AI" ], "tags_highlighted": [], "top_company": false, "isHiring": false, "nonprofit": false, "batch": "Summer 2023", "status": "Active", "industries": [ "B2B" ], "regions": [ "United States of America", "America / Canada", "Remote", "Fully Remote" ], "stage": "Early", "app_video_public": false, "demo_day_video_public": false, "app_answers": null, "question_answers": null, "url": "https://www.ycombinator.com/companies/opensteer", "api": "https://devasheeshg.github.io/yc-api/batches/summer-2023/opensteer.json", "year_founded": 2023, "city": "Los Angeles", "country": "US", "linkedin_url": "", "x_url": "", "fb_url": null, "cb_url": null, "github_url": null, "logo_url": "https://bookface-images.s3.amazonaws.com/small_logos/38b0ef5314d328b72aff9a8d9829350f181b5ace.png", "app_video_url": null, "dday_video_url": null, "company_photos": [], "primary_partner": { "name": "Aaron Epstein", "url": "https://www.ycombinator.com/people/aaron-epstein" }, "founders": [ { "user_id": 1369366, "full_name": "Tim Jang", "title": "Founder", "founder_bio": "Founder at Opensteer. Previously at Block.", "is_active": true, "linkedin_url": "https://linkedin.com/in/timjang3", "x_url": "", "avatar_thumb_url": "https://bookface-images.s3.us-west-2.amazonaws.com/avatars/9fa76a17acafebcc366921a150dbeeceb5469bd0.jpg" } ], "jobs": [], "news": [], "launches": [ { "id": 102076, "title": "Opensteer - Specialized agents that learn and run your business", "tagline": "A team of AI agents that run your business", "body": "Opensteer creates specialized AI agents that learn and use your tools, workflows, rules, and custom processes to become experts at one specific job.\n\nEach agent runs in a secure cloud workspace with its own browser, tools, memory, schedules, rules, and approval gates. Instead of one generic assistant trying to do everything, Opensteer agents are trained around one job and the way your company actually gets it done.\n\nWith a team of specialized agents, you essentially have a suite of AI employees that help run and accelerate your business 24/7.\n\n\n\n**The Problem**\n\nAgents today are too generic. They get a long prompt, a huge context window, and every tool loaded at once.\n\nBut real company workflows are specific and messy. Sales follow-up happens across a CRM, Gmail, LinkedIn, and Slack. Support triage may require tickets, docs, dashboards, and manager approvals. Ops work often lives inside browser-only portals with no API.\n\nEach job needs its own rules, context, workflows, and guardrails. With one generic agent, those rules get blended together, important context gets lost, and the quality of the work drops.\n\n**The Solution**\n\nOpensteer turns a generic agent into a specialized worker by giving it a custom harness for one job.\n\nA harness is a filesystem that teaches an agent how to do a workflow well. It contains the agent's operating playbook, memory, tools, schedules, browser access, deterministic code, approval rules, and integrations for that specific job.\n\nSkills tell the agent how to think, make tradeoffs, and follow your company's rules. Actions and scripts handle the parts that should run the same way every time: API calls, browser automation, parsing, validation, database updates, notifications, and scheduled routines.\n\nYou decide where each workflow sits on the autonomy spectrum: deterministic code, agent judgment, human approval, or a mix of all three.\n\nBecause harnesses can use APIs, first-party MCPs, CLIs, custom code, and browser automation, Opensteer agents can integrate with anything your team already uses.\n\n**How It Works**\n\n1\\. Describe the job you want the agent to own.\n\n2\\. Opensteer creates a specialized harness for that workflow.\n\n3\\. Connect the apps, browser profiles, tools, APIs, and data it needs.\n\n4\\. Add rules, approvals, schedules, and deterministic actions.\n\n5\\. Run it in the cloud, then check in or steer it from Opensteer, Claude Code, or Codex.\n\n**Our Ask**\n\nCreate your own custom, specialized agents today at Opensteer or book a call here: [**https://cal.com/tim-opensteer/30min**](https://cal.com/tim-opensteer/30min)\n\nWebsite: [**https://opensteer.com**](https://opensteer.com)\n\nEmail: [**tim@opensteer.com**](mailto:tim@opensteer.com)", "url": "https://www.ycombinator.com/launches/QYO-opensteer-specialized-agents-that-learn-and-run-your-business", "votes": 17, "created_at": "2026-05-28T07:11:36.274Z" }, { "id": 95163, "title": "Oversteer - The Most Reliable Browser Agent for Any Web Task", "tagline": "Turn any website into a reusable API", "body": "**_TL;DR_** Oversteer is building browser agents to automate web tasks and turn them into reusable, deterministic APIs. You can integrate the API into any system and automate/integrate any website or portal. \n\nhttps://www.youtube.com/watch?v=IZCy2r-rs7s\n\n# **What’s the insight?**\n\nAcross industries, teams still outsource repetitive web tasks — logging into portals, downloading files, submitting forms, extracting data — to human labor, spending millions of human labor hours. Traditional browser automation tools exist, but they’re **unreliable, expensive, and non-deterministic**. Companies and users want something that just works - a reliable agent that can execute their workflow end-to-end deterministically that works correctly every time, and actually replace human work. \n\n# **What did we build?**\n\nOversteer is an **AI browser agent** that automates any website or web portal from a simple prompt. It generates **deterministic scripts** that run like APIs — **reliable, reusable, and self-healing** when websites change. Because they don’t rely on LLMs at runtime, they’re **fast, consistent, and cheaper** than traditional browser agents.\n\n# **How it works:**\n\n**Step 1:** Understands and records a user’s web workflow from just a prompt.\\\n**Step 2:** Generates a deterministic automation script from the workflow.\\\n**Step 3:** Handles anti-bot detection, CAPTCHA solving, and automatic login with 2FA.\\\n**Step 4:** Runs as a reusable API and connects to systems like Zapier, n8n, and app builders like Lovable, v0.\\\n**Step 5:** Automatically adapts when websites update, ensuring the automation keeps working.\n\n---\n\n# **The Team**\n\n**Tim Jang** – Founder. Previously built multiple consumer apps with millions of users. Developed large-scale web automation systems to extract real-time data on millions of consumer products, and is now bringing that technology into a product that anyone can use.\n\n---\n\n# **Our Ask (and Offer)**\n\nTry it yourself for free: https://oversteer.dev\n\nIntro: We’re looking to connect with companies that spend hours doing repetitive web tasks — in **procurement, logistics, healthcare, accounting, finance**, or any industry with heavy manual computer work.\n\nOr if you’re someone who wants to integrate websites into automation workflow tools like n8n, Zapier, or app builders like Lovable, we’d love to talk.\n\nFeel free to email me at [**tim@oversteer.dev**](mailto:tim@oversteer.dev)\n\nor book a meeting here: https://cal.com/tim-jang-r1jrml/30min", "url": "https://www.ycombinator.com/launches/Okt-oversteer-the-most-reliable-browser-agent-for-any-web-task", "votes": 15, "created_at": "2025-11-05T02:17:41.406Z" } ] }, { "id": 29296, "name": "Simplex", "slug": "simplex", "former_names": [ "Simplex", "Pansimulate" ], "small_logo_thumb_url": "https://bookface-images.s3.amazonaws.com/small_logos/556d5a66ef0386d859db3ad9f500ceeadf756032.png", "website": "https://simplex.sh", "all_locations": "San Francisco, CA, USA", "long_description": "Simplex builds AI agents to handle provider enrollment end-to-end. We automate pulling provider data from CAQH, filling out enrollment forms + uploading licenses via payor portals/email/phone, updating applications when returned for corrections, and performing effective date follow-ups until a final determination is made, all autonomously.", "one_liner": "AI agents to enroll providers with payors", "team_size": 2, "industry": "B2B", "subindustry": "B2B -> Engineering, Product and Design", "launched_at": 1723709594, "tags": [ "Machine Learning", "Robotic Process Automation", "B2B", "AI" ], "tags_highlighted": [], "top_company": false, "isHiring": false, "nonprofit": false, "batch": "Summer 2024", "status": "Active", "industries": [ "B2B", "Engineering, Product and Design" ], "regions": [ "United States of America", "America / Canada" ], "stage": "Early", "app_video_public": false, "demo_day_video_public": false, "app_answers": null, "question_answers": null, "url": "https://www.ycombinator.com/companies/simplex", "api": "https://devasheeshg.github.io/yc-api/batches/summer-2024/simplex.json", "year_founded": 2024, "city": "San Francisco", "country": "US", "linkedin_url": "https://www.linkedin.com/company/usesimplex", "x_url": "https://x.com/usesimplex", "fb_url": null, "cb_url": null, "github_url": null, "logo_url": "https://bookface-images.s3.amazonaws.com/small_logos/556d5a66ef0386d859db3ad9f500ceeadf756032.png", "app_video_url": null, "dday_video_url": null, "company_photos": [ "https://bookface-images.s3.us-west-2.amazonaws.com/attachments/023924cfc3ff92ee7ef40347a383d95811798be6.png" ], "primary_partner": { "name": "Aaron Epstein", "url": "https://www.ycombinator.com/people/aaron-epstein" }, "founders": [ { "user_id": 1588677, "full_name": "Shreya Karpoor", "title": "Founder/CEO", "founder_bio": "Shreya is the co-founder and CEO of Simplex. She holds a BS and MEng in Electrical Engineering and Computer Science from MIT. She previously built software at Tesla and researched dexterous robotic manipulation at MIT.", "is_active": true, "linkedin_url": "https://linkedin.com/in/shreya-karpoor", "x_url": "", "avatar_thumb_url": "https://bookface-images.s3.us-west-2.amazonaws.com/avatars/2ea7b8584bb0c1fd567102ce20a5cf745915ce09.jpg" }, { "user_id": 910356, "full_name": "Marco Nocito", "title": "Founder/CTO", "founder_bio": "Marco is the co-founder and CTO of Simplex. He holds a BS and MEng in Computer Science from MIT. He previously built machine learning models to generate synthetic data at Waymo and built data infrastructure tooling at Viam and Bloomberg.", "is_active": true, "linkedin_url": "https://linkedin.com/in/marco-nocito/", "x_url": "", "avatar_thumb_url": "https://bookface-images.s3.us-west-2.amazonaws.com/avatars/dba39276e5b4d649b475457eb63e059283e2a78e.jpg" } ], "jobs": [], "news": [], "launches": [ { "id": 90728, "title": "Simplex - enterprise-grade web agents", "tagline": "AI agent companies automate legacy portals using our SDKs.", "body": "Simplex builds enterprise-grade web agents that companies use to integrate with legacy portals. We handle everything from building the web agents for you to integrating directly with your existing software stack like forward deployed engineers.\n\nCustomers use Simplex in production right now to:\n\n* submit information from outbound calls into freight software\n* edit and download their customers’ invoices\n* fetch websites’ internal APIs\n* and more.\n\nhttps://youtu.be/97gX3I9PMZ0\n\n# The Problem\n\nLegacy portals aren’t well suited for web agent automation — portals are unintuitive to navigate and functionality is often broken.  Existing web agents like Operator lack fine-grained control over agent behavior and can’t consistently perform these critical legacy workflows.\n\nA fully autonomous AI might fail to reach the desired page, get stuck pressing a broken button, or mistype critical information like an invoice amount or a shipping address. Without control over agent behavior and clear observability during workflows, web agents are blocked from being used during production in these critical scenarios.\n\n# The Solution\n\nSimplex is the production-grade web agent platform that enables AI agents to reliably do work on legacy portals at scale. You can see a live demo of Simplex below – the demo is at 1x speed.\n\nhttps://youtu.be/4VSexxYptV4\n\nWe’ve engineered Simplex to give AI agent developers control, transparency, and reliability. With Simplex you can:\n\n* easily **integrate with your existing stack** – pass the results from LLM calls, databases, your existing agents into Simplex\n* **cache element actions** for speed and consistent performance\n* build **strong evals** to stress-test your workflows at scale and emulate production conditions\n* scale up to **100s of concurrent browser sessions** while handling CAPTCHAs, 2FA, and website authentication on your customers’ portals\n* ensure **no incorrect actions** are taken — if you prompt our agents to click on elements that don’t exist on the page, we'll raise an exception\n* debug with shareable **session replays** and **detailed agent logs**\n\n# The Team\n\n[Shreya Karpoor](https://www.linkedin.com/in/shreya-karpoor/) and [Marco Nocito](https://www.linkedin.com/in/marco-nocito/) are MIT grads who’ve spent years doing computer vision work at Waymo, Tesla, and within MIT labs. They’re excited to use their expertise to bring reliable web agents into widespread production.\n\n# \n\n# Our Ask\n\n* [Book a call here](https://cal.com/marco-nocito-kon7ng/30min) if you want to request API access.\n* [Visit our website here](https://www.simplex.sh/) to see a live demo of Simplex.", "url": "https://www.ycombinator.com/launches/NbM-simplex-enterprise-grade-web-agents", "votes": 17, "created_at": "2025-05-27T14:42:32.436Z" }, { "id": 87190, "title": "Simplex - Web automation for developers", "tagline": "API to automate manual workflows on the web", "body": "**TL;DR:**\n\n* Simplex is an API that allows you to automate manual workflows on the web. For example, a customer uses us to autonomously purchase materials from Amazon, poll their email inbox for tracking information, then update their internal documentation with that tracking info every six hours.\n* Book [time with us here](https://calendly.com/simplex-interest/marco-chat) if we can help you automate your business.\n\nHi everyone, we’re [Shreya](https://www.linkedin.com/in/shreya-karpoor/) and [Marco](https://www.linkedin.com/in/marco-nocito/), two MIT grads building Simplex.\n\nTeams are often burdened by manual tasks that have to be repeated thousands of times per month — for example, populating spreadsheets with information from recent emails. From our experience, we found that asking AI agents to solve these problems just didn’t work. They often didn’t follow the same path when we asked them to do the same thing twice, and we couldn’t control any of the actions they took when they made a mistake.\n\nSo we built Simplex to inject intelligence into repeatable, known workflows to help you automate your manual tasks.\n\nWith Simplex, writing a function to pull tracking information from your email is as easy as this:\n\n![uploaded image](/media/?type=post&id=87190&key=user_uploads/910356/27f7171e-7a04-408e-84f9-d7596b9a2e5b)\n\n**Watch a quick demo here:**\n\n\n\nA few reasons customers love Simplex:\n\n* **Straight-forward, reliable intelligence:** We break up your web workflows into smaller, more atomic actions (click, type, extract) to enforce while allowing you to leverage the intelligence of AI.\n* **Extract anything on the web:** We’ve built our own vision system to more accurately extract small icons, text, images, or even entire sections of a page.\n* **Run on any site, even internal tooling:** We store your credentials securely to handle website authentication, meaning you can use Simplex on any website or private tooling you have access to.\n* **No need to deal with browser infrastructure yourself:** We handle all the infrastructure, hosting, and scheduling, letting you run these workflows asynchronously at scale.\n\nPeople have requested Simplex to:\n\n* **act as procurement software:** automatically order parts and update internal spreadsheets with delivery information across multiple vendors (Amazon, DigiKey, etc.)\n* **scrape financial information for their hedge fund clients:** pull daily headlines and perform sentiment analysis across hundreds of financial websites for quantitative analysis research\n* **autonomously update their internal database of products:** aggregate manufacturer marketing information for over 2,000,000+ retail products\n* **do the work of an entire data collection team:** collect grant and award information from over 700 government websites\n* and more…\n\n## Our Ask\n\n* We’re looking to onboard teams, especially those that are burdened by manual processes or need to scrape hundreds of thousands of websites. Book a [meeting with us](https://calendly.com/simplex-interest/marco-chat) for a demo and API access. We’re happy to write your first automation for you.", "url": "https://www.ycombinator.com/launches/MgI-simplex-web-automation-for-developers", "votes": 15, "created_at": "2025-01-27T15:36:17.454Z" }, { "id": 83077, "title": "Simplex: on-demand photorealistic vision datasets", "tagline": "Receive millions of high quality images and labels for training your AI models.", "body": "**TL:DR;** Simplex creates photorealistic vision datasets rendered from 3D scenes for AI model training. Submit a request [on our website](https://simplex.sh/#data-request-form) to receive high-quality data and labels.\n\n![uploaded image](/media/?type=post&id=83077&key=user_uploads/1588677/826c7bea-8936-451d-97ff-895bc8ca6c38)\n\n**Data request for the above sample:** _“Generate images and labels of a home kitchen with household objects on a center table. I need a variety of household objects in a variety of lighting conditions. Our desired labels are semantic segmentation and depth maps.”_\n\nHi everyone, we’re [**Shreya**](https://www.linkedin.com/in/shreya-karpoor/) and [**Marco**](https://www.linkedin.com/in/marco-nocito/), two MIT grads building Simplex.\n\nCollecting vision data for model training is time-consuming, costly, and often unsafe. Shreya spent over 200 hours physically operating a robot to collect image training data during her research at MIT. Marco worked on machine learning for synthetic data at Waymo to solve this exact problem.\n\nWe realized data scarcity wasn’t just an issue in robotics – it affects any company training vision models. When fine-tuning foundation models or building a new dataset from scratch, teams must curate existing data or label and collect data themselves. \n\nWe resolve the data scarcity problem by generating photorealistic ground truth labeled datasets for **any scenario**. We can generate **millions of varied images** **from 3D scenes** using our physics engine pipeline.\n\nHere’s how you’d use Simplex:\n\n1. Fill out our data request form [here](https://simplex.sh/#data-request-form) – it takes less than a minute. \n2. Give us feedback on a few sample image/label pairs that we generate. Repeat if necessary.\n3. Once you’re satisfied, download your complete dataset.\n\nWe support semantic segmentation, captions, simulated LiDAR, depth maps, and bounding boxes. You can generate large volumes of randomized scenes or provide a CAD/phone scan model for more specific scenes.\n\n# **Our Ask**\n\n* If you or someone you know needs vision data, fill out our 30-second data request [form.](https://www.simplex.sh/#data-request-form) We’re taking a limited number of early customers.\n* If you have a more complicated request or would otherwise like to contact us, email [shreya@simplex.sh](https://mailto:shreya@simplex.sh).\n\n# **The Team**\n\n![uploaded image](/media/?type=post&id=83077&key=user_uploads/1588677/d2e5778c-86c2-4c7b-bc75-3c73aefff883)\n\n[**Shreya**](https://www.linkedin.com/in/shreya-karpoor/): Computer science (BS and MEng) at MIT, software engineer at Tesla and Viam. Built simulation pipelines for locomotion and dexterous manipulation research at MIT. \\\n[**Marco**](https://www.linkedin.com/in/marco-nocito/): Computer science (BS and MEng) at MIT, software engineer at Waymo, Bloomberg, and Viam. Built machine learning models to generate synthetic data at Waymo.", "url": "https://www.ycombinator.com/launches/Lbx-simplex-on-demand-photorealistic-vision-datasets", "votes": 407, "created_at": "2024-08-15T14:29:52.537Z" } ] }, { "id": 29753, "name": "Autumn Labs", "slug": "autumn-labs", "former_names": [], "small_logo_thumb_url": "https://bookface-images.s3.amazonaws.com/small_logos/5df469af020f91ac0b4d19796abb3d31c3518288.png", "website": "https://www.autumnlabs.io/", "all_locations": "San Francisco, CA, USA", "long_description": "Autumn Labs delivers a developer-friendly platform for monitoring and managing modern manufacturing lines. Seamlessly integrating with robotic stations—automated test stations, robotic assembly cells, and industrial arms—it ensures full traceability, live data monitoring, and secure data transport. We prioritize streamlining factory operations, boosting production quality, and preventing supply chain disruptions, all while offering an effortless onboarding experience for engineers and manufacturers.", "one_liner": "Datadog for industrial robots", "team_size": 3, "industry": "Industrials", "subindustry": "Industrials -> Manufacturing and Robotics", "launched_at": 1724817075, "tags": [ "Hardware", "Robotic Process Automation", "SaaS", "Robotics", "Manufacturing" ], "tags_highlighted": [], "top_company": false, "isHiring": false, "nonprofit": false, "batch": "Summer 2024", "status": "Active", "industries": [ "Industrials", "Manufacturing and Robotics" ], "regions": [ "United States of America", "America / Canada", "Remote", "Partly Remote" ], "stage": "Early", "app_video_public": false, "demo_day_video_public": false, "app_answers": null, "question_answers": null, "url": "https://www.ycombinator.com/companies/autumn-labs", "api": "https://devasheeshg.github.io/yc-api/batches/summer-2024/autumn-labs.json", "year_founded": 2023, "city": "San Francisco", "country": "US", "linkedin_url": "", "x_url": "", "fb_url": null, "cb_url": null, "github_url": null, "logo_url": "https://bookface-images.s3.amazonaws.com/small_logos/5df469af020f91ac0b4d19796abb3d31c3518288.png", "app_video_url": null, "dday_video_url": null, "company_photos": [], "primary_partner": { "name": "Harj Taggar", "url": "https://www.ycombinator.com/people/harj-taggar" }, "founders": [ { "user_id": 991674, "full_name": "Prathik Muppidi", "title": "Founder/CEO", "founder_bio": "University of Waterloo Mechatronics Engineering, ex-Google Engineer", "is_active": true, "linkedin_url": "https://linkedin.com/in/prathikmuppidi", "x_url": "", "avatar_thumb_url": "https://bookface-images.s3.us-west-2.amazonaws.com/avatars/ae728870bbb8d337d10dc9aecf09f3058982fe3b.jpg" }, { "user_id": 1250574, "full_name": "Harsh Bhatt", "title": "Founder/CTO", "founder_bio": "Ex. Apple | Mechatronics Engineering @ University of Waterloo", "is_active": true, "linkedin_url": "https://linkedin.com/in/harshmbhatt", "x_url": "", "avatar_thumb_url": "https://bookface-images.s3.us-west-2.amazonaws.com/avatars/f9cfe0f35ded2657954fedd4b377c34d61c80e6c.jpg" }, { "user_id": 309960, "full_name": "Sam Monga", "title": "Founder/CTO", "founder_bio": "Hardware software engineer", "is_active": true, "linkedin_url": "https://linkedin.com/in/sammonga", "x_url": "", "avatar_thumb_url": "https://bookface-images.s3.us-west-2.amazonaws.com/avatars/ba15caa4ff5b343488e5acfc1fd1063b92da206c.jpg" } ], "jobs": [], "news": [], "launches": [ { "id": 83709, "title": "Autumn Labs - Monitoring platform for industrial robots and manufacturing", "tagline": "Detect machine downtime, process quality, and factory issues — in real time. ", "body": "**tl;dr:** [Autumn Labs](https://www.autumnlabs.io/) delivers a unified platform to monitor and manage modern factories. Our SDK seamlessly integrates with manufacturing systems (test stations, assembly robots, industrial arms, and manual stations) to provide real-time monitoring, control, and production data management. The result? Lower operation costs, higher product quality, and a smoother production line.\n\n🍁 = 🤖 + 🏭 + 📊\n\n### **Problem - Advanced Machines, Outdated Infrastructure**\n\nThe adoption of robots in manufacturing is exploding, especially to keep up with producing increasingly more complex products and shortages in skilled labor. However, the infrastructure in which these robots would be deployed is still years behind.\n\nWithout effective systems, manufacturing physical products might seem like an insurmountable black box, especially when they are made in factories on another continent where critical information is only reported via spreadsheets every other week.\n\nThe past: [_src_](https://www.youtube.com/watch?app=desktop&v=xqPfY8S5HJ4)\n\n![uploaded image](/media/?type=post&id=83709&key=user_uploads/991674/84c0d0be-c84c-4872-84f8-20aa591d28a6)\n\nToday: [_src_](https://www.youtube.com/watch?app=desktop&v=luFGI13Mv8o)\n\n![uploaded image](/media/?type=post&id=83709&key=user_uploads/991674/2da6a941-47af-48e3-aaf4-4ddee5351437)\n\nBut the manufacturing data is always lagging and still looks like this:\n\n![uploaded image](/media/?type=post&id=83709&key=user_uploads/991674/5472966f-362b-483b-9117-37b4730a3676)\n\n### **Solution - Seamless Integration, Real-time insights**\n\nAutumn Labs provides an SDK that’s easy to integrate into any robotic architecture, enabling seamless monitoring and management of machines and, ultimately, the factory in which they reside. This way, robotic integrators and developers have a standard platform where they pull and push critical data that’s specific to their machine’s tasks. Both factory operations and engineering teams are able to get actionable data they need to improve production utilizations and process quality.\n\n_A Unified Platform for Hybrid Factories:_\n\nOur platform is built for hybrid factory lines, too, allowing for seamless integration of manual human assembly stations. Factory personnel can now get a holistic view of your factory's performance—robotic or not.\n\n![uploaded image](/media/?type=post&id=83709&key=user_uploads/991674/8d55d085-5141-4c8e-a11f-d4751a050099)\n\nWith Autumn Labs, you can track, triage, and push fixes for:\n\n1. Robotic assembly stations/cells\n2. Automated test stations\n3. Palletizing robots\n4. Manual assembly stations\n\n### **Traction - Real Results, Real Impact:**\n\nAutumn Labs is already monitoring and managing 50+ automated factory stations in 4 different consumer electronics factory lines across multiple companies and contract manufacturers. There’s a good chance that a product you use today has gone through our platform.\n\nWhy our customers love us:\n\n1. Clear insights into Manufacturing Value Added (MVA) costs\n2. Increased machine utilization\n3. Rapid improvements in production quality\n4. Enhanced product development cycles\n\n### **Meet the Founders**\n\nWe’re [Sam](https://www.linkedin.com/in/sammonga/), [Harsh](https://www.linkedin.com/in/harshmbhatt/), and [Pra](https://www.linkedin.com/in/prathikmuppidi/)[thik](https://www.linkedin.com/in/harshmbhatt/) — Mechatronics engineers from the University of Waterloo with decades of combined experience at Apple (iPhone), Google (Nest), and Meta (Oculus). After countless sleepless nights fixing robotic assembly lines and test stations, we decided it was time to create the ultimate tool for manufacturing. We’re building the solution we wish we had back then.\n\n![uploaded image](/media/?type=post&id=83709&key=user_uploads/991674/4ecd6fa2-6352-46ae-a602-f69219f0fe93)\n\n### **Our Ask - Let's Transform Manufacturing Together**\n\nKnow someone struggling to get real-time data from robots, test stations, or factories? Let’s talk. Reach us at [founders@autumnlabs.io](mailto:founders@autumnlabs.io)", "url": "https://www.ycombinator.com/launches/Lm9-autumn-labs-monitoring-platform-for-industrial-robots-and-manufacturing", "votes": 35, "created_at": "2024-09-03T16:30:31.314Z" } ] }, { "id": 29847, "name": "Bucket Robotics", "slug": "bucket-robotics", "former_names": [], "small_logo_thumb_url": "https://bookface-images.s3.amazonaws.com/small_logos/55cc697d9f14eca5a5d45772ac14d97a83df16ae.png", "website": "https://bucket.bot", "all_locations": "San Francisco, CA, USA", "long_description": "Bucket Robotics builds deployable computer vision systems for manufacturing — without labeling, long pilots, or fragile rules.\r\nWe turn CAD and sample data into production-ready vision models that run on existing cameras and edge hardware. No manual labeling. No cold-start problem. Models deploy in minutes and adapt as parts, defects, and lines change.\r\n\r\nOur roots are in self-driving cars (Argo AI, Uber ATG, Stack AV), where we learned how to ship perception systems that operators trust in messy, real-world environments. We’re applying that playbook to factories: robust sensing, fast iteration, and tooling engineers actually enjoy using.\r\n\r\nManufacturers struggle with inspection because parts are too variable for rules-based vision and too expensive to inspect by hand. Legacy systems are rigid, hardware-locked, and slow to customize. We take the opposite approach: software-first vision that fits into existing automation and scales across SKUs, facilities, and workflows.\r\n\r\nAmerican Manufacturing is in the middle of a $700B automation push — but quality is still a trust problem. Bucket Robotics is building the vision infrastructure that lets factories automate inspection with confidence.", "one_liner": "Defect detection for manufacturing built from CAD and synthetic data.", "team_size": 8, "industry": "Industrials", "subindustry": "Industrials -> Manufacturing and Robotics", "launched_at": 1719001491, "tags": [ "Robotic Process Automation", "Robotics", "Computer Vision", "Manufacturing" ], "tags_highlighted": [], "top_company": false, "isHiring": true, "nonprofit": false, "batch": "Summer 2024", "status": "Active", "industries": [ "Industrials", "Manufacturing and Robotics" ], "regions": [ "United States of America", "America / Canada" ], "stage": "Early", "app_video_public": false, "demo_day_video_public": false, "app_answers": null, "question_answers": [ { "question": "What is your long-term vision? If you truly succeed, what will be different about the world?", "answer": "If we succeed, quality inspection stops being a human bottleneck. Manufacturing quality becomes software—deployed from CAD, scaled instantly across factories, and verified automatically. The world builds complex physical products faster, with less waste, and with machine-verifiable trust." } ], "url": "https://www.ycombinator.com/companies/bucket-robotics", "api": "https://devasheeshg.github.io/yc-api/batches/summer-2024/bucket-robotics.json", "year_founded": 2024, "city": "San Francisco", "country": "US", "linkedin_url": "https://www.linkedin.com/company/bucketrobotics/", "x_url": "https://x.com/BucketRobotics", "fb_url": "https://www.facebook.com/profile.php?id=61578008815737", "cb_url": null, "github_url": "https://github.com/bucket-robotics", "logo_url": "https://bookface-images.s3.amazonaws.com/small_logos/55cc697d9f14eca5a5d45772ac14d97a83df16ae.png", "app_video_url": null, "dday_video_url": null, "company_photos": [ "https://bookface-images.s3.us-west-2.amazonaws.com/attachments/72627bb1c990f52131cd9089f7589fc3dc47457f.png" ], "primary_partner": { "name": "Diana Hu", "url": "https://www.ycombinator.com/people/diana-hu" }, "founders": [ { "user_id": 1918803, "full_name": "Matt Puchalski", "title": "Founder/CEO", "founder_bio": "Matt is the co-founder and CEO of Bucket Robotics, building CAD-driven computer vision for manufacturing inspection and assembly verification. Previously, he worked on autonomous vehicle integration and deployment at Argo AI, Latitude AI, and Stack AV, helping bring self-driving vehicles onto public roads in the United States and Germany.", "is_active": true, "linkedin_url": "https://www.linkedin.com/in/matt-puchalski", "x_url": "https://x.com/lasermatts", "avatar_thumb_url": "https://bookface-images.s3.us-west-2.amazonaws.com/avatars/96e15f894eb00281539dee8a84cf516b581aac67.jpg" } ], "jobs": [ { "id": 73634, "title": "ML Engineer", "url": "https://www.ycombinator.com/companies/bucket-robotics/jobs/wamcLUR-ml-engineer", "location": "San Francisco, CA, US", "type": "Full-time", "role": "Engineering", "role_type": "Machine learning", "salary_range": "$120K - $165K", "equity_range": "0.70% - 2.80%", "experience": "3+ years", "visa": "US citizen/visa only", "skills": [ "ML", "Python", "Machine Learning", "CUDA" ] }, { "id": 95125, "title": "Senior Full Stack Engineer", "url": "https://www.ycombinator.com/companies/bucket-robotics/jobs/6pT0gS3-senior-full-stack-engineer", "location": "San Francisco, CA, US", "type": "Full-time", "role": "Engineering", "role_type": "Full stack", "salary_range": "$160K - $220K", "equity_range": "1.00% - 3.00%", "experience": "6+ years", "visa": "US citizen/visa only", "skills": [ "Data Modeling" ] }, { "id": 95797, "title": "Forward Deploy Engineer", "url": "https://www.ycombinator.com/companies/bucket-robotics/jobs/va6hXZz-forward-deploy-engineer", "location": "San Francisco, CA, US", "type": "Full-time", "role": "Engineering", "role_type": "Robotics", "salary_range": "$100K - $120K", "equity_range": "0.25% - 1.20%", "experience": "1+ years", "visa": "US citizen/visa only", "skills": [ "Python" ] } ], "news": [ { "title": "Synthetic Data Comes of Age: How Bucket Robotics is Changing the Economics of Machine Vision - MVPro Media", "url": "https://mvpromedia.com/synthetic-data-comes-of-age-how-bucket-robotics-is-changing-the-economics-of-machine-vision/", "date": "Jul 21, 2026" }, { "title": "How YC-backed Bucket Robotics survived its first CES | TechCrunch", "url": "https://techcrunch.com/2026/01/18/how-yc-backed-bucket-robotics-survived-its-first-ces/", "date": "Jan 18, 2026" } ], "launches": [ { "id": 82809, "title": "Bucket Robotics 🤖 Vision for Manufacturing", "tagline": "Go from CAD model to defect detection system in the blink of an eye.", "body": "# ☕ **_TL;DR_**\n\n[_Bucket Robotics_](https://bucket.bot) makes defect detection models way faster and easier to deploy. CAD goes in — detection model comes out.\n\n![uploaded image](/media/?type=post&id=82809&key=user_uploads/1918803/d49ec265-a7cd-4a9c-ad32-62b45df8c825)\n\n# **👥** **The Team**\n\n### Hi everyone, we’re [Matt](https://www.linkedin.com/in/matt-puchalski/) and [Steph](https://www.linkedin.com/in/stephan-wolski/), and we’re Bucket Robotics!\n\n![uploaded image](/media/?type=post&id=82809&key=user_uploads/1918803/a2825a9a-9dd1-4c95-8aa4-46ba2826efae)\n\n# **📉 The Problem**\n\nDefects are a universal challenge in manufacturing, particularly with methods like injection molding, die casting, and 3D printing. Factories using these methods frequently retool their lines for different products making it difficult and costly to gather enough data to train defect detection models.\n\nManufacturers strive to ensure that only the highest quality products reach their customers. Avoiding the shipment of defective products not only protects their reputation but also safeguards expensive machinery from damage, maximizes yield by leveraging advanced analytics, and prevents the progression of faulty parts through the production line, which can lead to compounded waste and increased costs.\n\nA 2017 study estimates that [yield losses account for 44% of sheet metal used](https://www.sciencedirect.com/science/article/abs/pii/S092401361730211X) in the production of passenger vehicles — and we want to bring that number down, saving money and the environment.\n\nTraditionally, defect detection comes in two flavors:\n\n1. Pay a person to watch your line.\n2. Train a vision model the old-fashioned way — wait for defective parts to come out and add your sample to the dataset.\n\nPaying people to check parts is inefficient and a miserable job. Collecting enough labeled data samples to train a model is impractical for rapidly retooled outputs, and the labeling is expensive. This often leaves manufacturers with a critical gap in quality assurance, leading to increased scrap, downtime, and potentially unhappy customers with defective parts.\n\n## **📈 The Solution**\n\nWe help eliminate the need for inefficient human monitoring and outdated defect detection by providing a system where we generate defect detection models using CAD files—something every part designed for modern manufacturing has. Our models can run on an edge computing device as small as a Jetson Orin Nano and be continually refined with real-world data.\n\n### **_Here’s why this is a game-changer:_**\n\n* **Early Deployment:** Your defect detector can arrive at your manufacturing line before you’ve manufactured any parts. This means you can catch defects right from the get-go.\n* **Continuous Improvement:** As defects start to appear, we can automatically refine and improve the model, ensuring your system remains top-notch.\n* **Classification**: The system tells you how the part is defective, allowing you to dial in your tooling configuration and maintenance cycle—e.g. when it's time to create a new mold.\n\nCheck out a video on our process here: [Bucket Robotics | Defect Detection](https://youtu.be/U_bJxaH_U9U)\n\n# **🙏How You Can Help**\n\nWe are seeking connections and collaborations to drive our mission forward. If you or anyone in your network is interested in learning more or helping out, we want to hear from you! We’re specifically looking to connect with:\n\n* Quality Assurance and Factory Automation leaders\n* Companies and brands in high-volume and precision manufacturing\n* Experts in the manufacturing and defect detection industries\n\nIf that’s you or someone you know, [please get in touch!](https://bucket.bot)", "url": "https://www.ycombinator.com/launches/LXd-bucket-robotics-vision-for-manufacturing", "votes": 77, "created_at": "2024-08-07T05:00:27.470Z" } ] }, { "id": 30174, "name": "Revise Robotics", "slug": "revise-robotics", "former_names": [], "small_logo_thumb_url": "https://bookface-images.s3.amazonaws.com/small_logos/f4e9ac60aa5b0583ce914c6dee2848e5b72c4b4f.png", "website": "https://www.reviserobotics.com/", "all_locations": "New York City, NY, USA", "long_description": "E-waste recycling facilities spend most of their time and effort sifting through discarded electronics, searching for valuable devices, refurbishing them, and reselling them in second-hand markets.\r\n\r\nOur first product is an AI-enabled robotic system that integrates with these facilities to assess and refurbish hundreds of laptops a day without any human intervention. The system tests laptops, wipes their hard drives, photographs them, and posts them on online marketplaces for resale, regardless of model, manufacturer, or operating system.", "one_liner": "Automating refurbishment of $1T in consumer electronics", "team_size": 3, "industry": "Industrials", "subindustry": "Industrials -> Manufacturing and Robotics", "launched_at": 1737068200, "tags": [ "Robotic Process Automation", "ClimateTech", "Electronics" ], "tags_highlighted": [], "top_company": false, "isHiring": true, "nonprofit": false, "batch": "Winter 2025", "status": "Active", "industries": [ "Industrials", "Manufacturing and Robotics" ], "regions": [ "United States of America", "America / Canada" ], "stage": "Early", "app_video_public": true, "demo_day_video_public": false, "app_answers": null, "question_answers": null, "url": "https://www.ycombinator.com/companies/revise-robotics", "api": "https://devasheeshg.github.io/yc-api/batches/winter-2025/revise-robotics.json", "year_founded": 2024, "city": "New York City", "country": "US", "linkedin_url": "https://www.linkedin.com/company/revise-robotics/", "x_url": "", "fb_url": null, "cb_url": null, "github_url": null, "logo_url": "https://bookface-images.s3.amazonaws.com/small_logos/f4e9ac60aa5b0583ce914c6dee2848e5b72c4b4f.png", "app_video_url": "https://bookface-images.s3.us-west-2.amazonaws.com/bf-public_profile/30227/app_video.mp4?X-Amz-Algorithm=AWS4-HMAC-SHA256&X-Amz-Credential=ASIAQC4NIECAODIUI2QA%2F20261020%2Fus-west-2%2Fs3%2Faws4_request&X-Amz-Date=20261020T022724Z&X-Amz-Expires=60&X-Amz-Security-Token=IQoJb3JpZ2luX2VjEHEaCXVzLXdlc3QtMiJGMEQCICdQyGY%2BbftNLxtKIYzFPxdUXttuiP0N09%2F%2BUfBHiBuhAiBOSl161zU1GSzXg078adXidT%2BI1mNTN00xIiFabrEuZSrlAwg6EAAaDDAwNjIwMTgxMTA3MiIMkex8rUdJQnYvqkb2KsIDZvHjdb5Nj1eMSDUiIIyyiwfJ8RyPFxer5o1XPlMS8x%2FtN9n%2F7PvS5Wd7oFPHi1qVxZAOUh%2BsJ%2B%2BXOPvWZ3Ae6ZZbt6%2BpSxxIqhHajGvQPEffXaUpP5qEh5zdOBvoNiFOEcL3U5kvAXPdFAeyEQUSC22Zz6qlwEbD4gyjAVIG69Bq2BqpO5gGA142dDrsMk9Fpvvrw4sUrmoFsqVRjQeNNnbJ6kGBUpNWGmGUg9xVdF1g6zjy5AOLq9gaZMWpGvdL2up4UE8tZcCtovhQX7LttuknA2oMDZbhb%2Bm1111BTyQwfpo0mzYMWvLSVFktebdbrXNkCCYHuwzkc%2FAYLLBCHcwr959rGhfijkmBq8zJ6L%2F3m2hFjzDHCnZYPOz0rFWk1O%2FdpaLpo4%2B83owIv8pdLkfxnSwz7jpV83o7ddFuQwi3GyYhgDV9NKraaZUPDf35OffhHBAa9LNH9KBtXE%2B9EJapzAmTZRHX0y1pAaJdqK7NnCs3JXj5MlJbG%2Fk7qe6hMwsDLiNtlbMQ85QzyJJHGU4y3dM2z2PBRjDSvRfVwapQjnLimR6h7fxLZHe1JssWKCLL3Y7dNZqEc8LFx7ZPmRWFMOj92tYGOqYB1OGPMkYIGSgzUnwJVB6Fm37ysfFBg0lS7An85D5284tv08ntQNXM%2B3NjFf0CnSg8DB6Uagn6Pf8Hz0vPosKjbkPeKHdUWICNI6ZBqrBsIZuFlEJamVl3d%2B5M5pi50RkfxaC0jUNRP0kzypvFoXoBWM5zGuVWthjEJhSlwkhKt8FAj4eChg9bDGTNeM9DkgbxKs0YI84QCmRK0pRtLAv0T11YvQ%2FTTg%3D%3D&X-Amz-SignedHeaders=host&X-Amz-Signature=a3315cd1ec85a06bb175d3d79563ec6a79b849bc96062b59f3ad085e840ea3d8", "dday_video_url": null, "company_photos": [], "primary_partner": { "name": "Garry Tan", "url": "https://www.ycombinator.com/people/garry-tan" }, "founders": [ { "user_id": 943950, "full_name": "Rupesh Jeyaram", "title": "Founder/CEO", "founder_bio": "Building AI-enabled robotic systems to refurbish and resell used electronic devices, starting with laptops. Wharton '24, Caltech '20", "is_active": true, "linkedin_url": "https://www.linkedin.com/in/rupesh-jeyaram/", "x_url": "", "avatar_thumb_url": "https://bookface-images.s3.us-west-2.amazonaws.com/avatars/e904201852f0dfce14feb1c345d0be02af322b85.jpg" }, { "user_id": 1985135, "full_name": "Antonio Monreal", "title": "Founder/CTO", "founder_bio": "Developing robots for electronics refurbishment", "is_active": true, "linkedin_url": "https://www.linkedin.com/in/antonio-monreal-71b0751b5/", "x_url": "", "avatar_thumb_url": "https://bookface-images.s3.us-west-2.amazonaws.com/avatars/d0a1cb5d67efed639fbb5772386daa39d0e6a022.jpg" } ], "jobs": [ { "id": 93937, "title": "Business Development Intern – Summer/Fall 2026", "url": "https://www.ycombinator.com/companies/revise-robotics/jobs/cNqu0M0-business-development-intern-summer-fall-2026", "location": "New York City", "type": "Full-time", "role": "Sales", "role_type": null, "salary_range": "$8K - $12K / monthly", "equity_range": "", "experience": "Any (new grads ok)", "visa": "Will sponsor", "skills": [] }, { "id": 93938, "title": "Engineering Intern — Summer/Fall 2026", "url": "https://www.ycombinator.com/companies/revise-robotics/jobs/peSRWPy-engineering-intern-summer-fall-2026", "location": "New York City", "type": "Full-time", "role": "Engineering", "role_type": "Full stack", "salary_range": "$8K - $12K / monthly", "equity_range": "", "experience": "Any (new grads ok)", "visa": "Will sponsor", "skills": [] } ], "news": [], "launches": [ { "id": 87007, "title": "Revise Robotics – Automated Electronics Refurbishment 🦾 💻", "tagline": "Our robot tests, refurbishes, and resells used electronics devices!", "body": "**At Revise Robotics, we’re building automated systems to refurbish and resell used electronics.**\n\n![uploaded image](/media/?type=post&id=87007&key=user_uploads/943950/79b9f19d-2610-4a84-b329-7b8624645181)\n\nEvery year, Americans discard more than **30 million computers**. Many of these devices find a second life on the resale market, extending their usefulness and reducing their environmental footprint.\n\nHowever, the process – inspection, data-wiping, reprogramming, and packaging – is **extremely** **tedious and time-consuming**, limiting how many devices can actually re-enter circulation.\n\nWe’re changing that. With robotics, computer vision, and AI, we’re breaking down these barriers and advancing a more circular electronics industry 💻\n\nWe just completed our first pilot project last month, and we are excited to be iterating and gearing up for our second pilot at the end of February!\n\n**Ask**: Introductions to e-waste recycling facilities or IT asset management/disposition companies. Please reach out if there’s someone you think we should speak to! [rupesh@reviserobotics.com](mailto:rupesh@reviserobotics.com) \n\n# **Full Demo Video**\n\n[**https://youtu.be/QMFTsSG-FyQ**](https://youtu.be/QMFTsSG-FyQ)\n\n# **Still Shot**\n\n![uploaded image](/media/?type=post&id=87007&key=user_uploads/943950/ca646ef9-8d31-4865-93cd-6b629a1612c3)\n\n", "url": "https://www.ycombinator.com/launches/MdL-revise-robotics-automated-electronics-refurbishment", "votes": 27, "created_at": "2025-01-21T16:43:12.194Z" } ] }, { "id": 30259, "name": "Asteroid", "slug": "asteroid", "former_names": [ "Entropy Labs" ], "small_logo_thumb_url": "https://bookface-images.s3.amazonaws.com/small_logos/4598dd293d773fa0a9a1c72940f2f2e3153dda35.png", "website": "https://asteroid.ai", "all_locations": "San Francisco, CA, USA", "long_description": "Asteroid is the AI-powered integration layer for healthcare. One connector for EHRs, payer portals, and legacy desktop systems, with 100+ pre-built integrations that go live in days, not months. Built for developers and operations teams, with integrations that maintain themselves.", "one_liner": "The integration layer for healthcare: EHRs, payer portals & more ", "team_size": 6, "industry": "B2B", "subindustry": "B2B", "launched_at": 1734639467, "tags": [ "Robotic Process Automation", "Health Tech", "B2B", "Automation", "AI" ], "tags_highlighted": [], "top_company": false, "isHiring": true, "nonprofit": false, "batch": "Winter 2025", "status": "Active", "industries": [ "B2B" ], "regions": [ "United States of America", "America / Canada" ], "stage": "Early", "app_video_public": false, "demo_day_video_public": false, "app_answers": null, "question_answers": null, "url": "https://www.ycombinator.com/companies/asteroid", "api": "https://devasheeshg.github.io/yc-api/batches/winter-2025/asteroid.json", "year_founded": 2024, "city": "San Francisco", "country": "US", "linkedin_url": "https://www.linkedin.com/company/asteroidai/", "x_url": "https://x.com/asteroid_inc", "fb_url": null, "cb_url": "https://www.crunchbase.com/organization/asteroid-9e1c", "github_url": "https://github.com/asteroidai", "logo_url": "https://bookface-images.s3.amazonaws.com/small_logos/4598dd293d773fa0a9a1c72940f2f2e3153dda35.png", "app_video_url": null, "dday_video_url": null, "company_photos": [ "https://bookface-images.s3.us-west-2.amazonaws.com/attachments/223c518cbae6486dd03a7709aace37e8071aae3d.png", "https://bookface-images.s3.us-west-2.amazonaws.com/attachments/5d89729aa0ccb3844b77004c5ef19da6ccbad2c0.png" ], "primary_partner": { "name": "Harj Taggar", "url": "https://www.ycombinator.com/people/harj-taggar" }, "founders": [ { "user_id": 1651915, "full_name": "David Mlcoch", "title": "Founder/CEO", "founder_bio": "Co-founder and CEO of Asteroid.\n\nI have a background in AI (Edinburgh University), autonomous agents (Co-founder and CTO of Trident AI), robotics (CTU, NTU, TAU) and space (EUSPA, SpaceKnow). I like sports, history and building things.\n", "is_active": true, "linkedin_url": "https://www.linkedin.com/in/davidmlcoch/", "x_url": "https://x.com/MlcochDavid", "avatar_thumb_url": "https://bookface-images.s3.us-west-2.amazonaws.com/avatars/a26cbf3a786ad8b0e9929f7d4d24e685714a3907.jpg" }, { "user_id": 1891505, "full_name": "Joe Hewett", "title": "Founder/CTO", "founder_bio": "Co-founder and CTO of Asteroid. \n\nPreviously cybersecurity at Netcraft, where I created novel methods using LLMs to map and disrupt hundreds of millions of dollars of fraud. I have published academic work on the use of LLMs in cybersecurity", "is_active": true, "linkedin_url": "https://www.linkedin.com/in/joehewett/", "x_url": "https://x.com/simulated_land", "avatar_thumb_url": "https://bookface-images.s3.us-west-2.amazonaws.com/avatars/425461b5427ba1aa60f5e01d3aceeb1696c688c4.jpg" } ], "jobs": [ { "id": 114163, "title": "Forward Deployed Engineer (Healthcare)", "url": "https://www.ycombinator.com/companies/asteroid/jobs/tYlxU3M-forward-deployed-engineer-healthcare", "location": "San Francisco, CA, US", "type": "Full-time", "role": "Engineering", "role_type": "AI engineer", "salary_range": "$120K - $160K", "equity_range": "0.20% - 0.35%", "experience": "1+ years", "visa": "US citizen/visa only", "skills": [ "LLMs", "AI Agents", "Playwright" ] } ], "news": [], "launches": [ { "id": 96061, "title": "Asteroid: AI Builder for complex Browser Agents", "tagline": "Build self-healing AI browser agents in minutes. Ops build, developers deploy, Asteroid self-improves.", "body": "### **TL;DR**\n\n**Asteroid** is the fastest way to build self-healing browser agents, no technical experience required. \\\n\\\nOur new **AI agent builder**, **Astro**, allows you to build **complex, multistep browser agents** by building an agent graph, writing prompts, adding tools, human in the loop and guardrails.\n\nThis makes Asteroid agents suitable for **mission-critical workflows**, and more than **25 businesses** already use them in healthcare, insurance, and beyond.\\\n\\\n**Astro** **is fully available, build your agent [here](https://platform.asteroid.ai/)!**\n\nhttps://youtu.be/ugY_qLTfmG8\n\n### **What is the problem?**\n\nStartups need to integrate with external portals (carriers, suppliers, healthcare) but can’t wait months for APIs.\n\nMany great browser-agent frameworks have emerged in the past year, but **non-technical teams still can’t build production-grade agents.**\n\nOperations leaders, PMs, and founders are the ones who **understand the workflows and edge cases** and _should_ be the builders. Relying on developers to constantly fix prompts for every new edge case isn’t sustainable.\n\n![uploaded image](/media/?type=post&id=96061&key=user_uploads/1651915/d73eee4d-bcc6-461f-953e-6fdd56374905)\n\n### **Solution: Astro - AI Builder for Complex Browser Agents**\n\nToday we’re launching **Astro**, our AI builder that lets anyone create **complex, self-healing browser agents in under 2 minutes**. Tell Astro what you want, for example _“Build an agent to pull patient data from this EHR”_, and it generates the full agent.\n\n![uploaded image](/media/?type=post&id=96061&key=user_uploads/1651915/53714daf-535a-450a-8a76-82eb70d43007)\n\n**Astro improves the agent using previous runs.** Behind the scenes, it writes code when needed, including Playwright scripts for deterministic steps, giving everyone code-level reliability.\n\n![uploaded image](/media/?type=post&id=96061&key=user_uploads/1651915/62627740-6f1e-47ce-a197-917e01ceade7)\n\nEvery part of the agent, including graphs, prompts, and model settings, is fully visible so you can make precise, surgical edits whenever needed.\n\n### **How does it work**\n\n1. **Build:** Non-technical teams create agents directly in the platform. Astro generates the prompts, agent graphs, and tools needed to handle even the most complex browser workflows.\n2. **Deploy:** Developers drop in a single code snippet to call the agent through Asteroid’s API/SDKs.\n3. **Supervise:** Monitor live sessions, review recordings, track performance, and trigger human-in-the-loop via Slack when needed.\n4. **Improve:** Non-technical teams update prompts and ship new agent versions instantly as workflows change.\n\n### **Ship your first agent today**\n\n**Astro** **is fully available, try it for free [here](https://platform.asteroid.ai/) or book a [demo](https://cal.com/asteroidai/15min) with a founder!**", "url": "https://www.ycombinator.com/launches/OzN-asteroid-ai-builder-for-complex-browser-agents", "votes": 228, "created_at": "2025-12-09T22:31:58.000Z" }, { "id": 88012, "title": "Asteroid: Browser Agents with Human Oversight", "tagline": "Browser workflow automation for regulated industries", "body": "**Asteroid makes browser automation seamless for regulated industries.** Not only developers but also non-technical teams can **build and run our browser agents with Asteroid**, turning 45-minute manual tasks into 2-minute automated processes.\n\nOur browser agents can navigate legacy portals, fill out forms, and extract data—all without brittle scripts, high costs, or maintenance headaches.\n\n![uploaded image](/media/?type=post&id=88012&key=user_uploads/1651915/50a0915f-b720-4fff-923b-0b84e6017098)\n\n### **The Problem**\n\nTraditional automation tools like RPA are too expensive for non-enterprise customers, **costing $10,000 per automation**. Meanwhile, scaleups especially in high-risk industries like insurance or healthcare lack the resources to maintain Selenium or Playwright scripts, as these frequently break due to UI changes and the automations need to be extremely reliable.\n\n### **The Solution**\n\nBrowser agents bridge this gap—offering the flexibility of scripting with the reliability of RPA. With 2025’s computer use models, **web automation is now 100x cheaper and more robust, unlocking thousands of new use cases.**\n\n### **How It Works**\n\n**Asteroid transforms any browser task into an agentic workflow.** Simply show Asteroid what to do—via a screen recording, SOP, or written instructions—and get an agentic workflow. You can then call it through our web UI or API.\n\nBehind the scenes, Asteroid spins up a secure, headless browser, handles authentication, executes workflows, and seamlessly integrates human review when needed—all without burdening your developers. We ensure the agent improves over time.\\\n🔗 [Watch Asteroid in action](https://www.youtube.com/watch?v=26uksP36xB4)\n\n![uploaded image](/media/?type=post&id=88012&key=user_uploads/1651915/fa0c7345-c244-4270-8c56-c905072f93ea)\n\n### **Highlighted Features**\n\n🔹 **API-First Automation** – Trigger a browser agent in a web UI or with a single API call—no need for complex scripts or RPA setups.\\\n🔹 **Human-in-the-Loop** – Automate only where it’s safe, with built-in escalation for manual review when needed.\\\n🔹 **Intelligent Guardrails** – Set rules like “Never click this button” or “Ask before submitting” to prevent errors.\\\n🔹 **Continuous Learning** – Asteroid evaluates every execution, adapting to website changes and improving over time using evaluations and backtesting.\n\n### **How We Got Here**\n\nWe’re both AI engineers who had to solve the same problem independently. Joe built AI systems that stopped $100M+ in fraud annually. David developed fraud investigation agents for fintechs.\n\nAfter years of building mission-critical automation, we learned that automation only works with intelligent guardrails, human oversight, and continuous learning.\n\nWe first built Asteroid as a developer tool for safe AI deployment—offering guardrails, human-in-the-loop workflows, and continuous evaluations. But we quickly realized:\n\n\\*\\*Browser agents need these safeguards even more.\\*\\*\\\n✅ **Guardrails** – \"Don’t click on this.\"\\\n✅ **Human oversight** – \"Ping me before you submit.\"\\\n✅ **Continuous learning** – \"The website changed—adapt.\"\n\nInstead of just developer tools, we’ve shifted to building browser agents with safety and reliability baked in!\n\n![uploaded image](/media/?type=post&id=88012&key=user_uploads/1651915/e9051949-8032-4fc5-951d-1fb8361ba3e1)\n\n### **Real-World Use Cases**\n\nWe’re working with customers to automate high-value workflows like:\n\n🔹 **Insurance Portal Automation** – Retrieve broker quotes across insurer portals—without manual input.\\\n🔹 **Supply Chain Scheduling** – Log into carrier platforms, retrieve GPS data, and schedule deliveries—seamlessly.\\\n🔹 **Expense Management** – Pull PDF bills from utility portals for expense tracking—without APIs.\n\n### **Who is Asteroid for?**\n\n🔹**Businesses with Heavy Manual Browser Workflows** – Automate large-scale data entry, form submissions, and repetitive portal interactions.\\\n🔹**Regulated Businesses using Legacy Systems** – Streamline automation for industries like insurance, healthcare and supply chain.\\\n🔹**AI Agent Startups** – Enable AI agents to reliably interact with legacy portals and automate browser tasks.\n\n### **Get Started**\n\n**Have a repetitive workflow you'd love to automate?** Let’s explore how Asteroid can help!\\\n🔹 [Schedule a demo](https://calendly.com/founders-asteroid-hhaf/30min?month=2025-02) to discuss your use case!\\\n🔹[Try our default agents](https://platform.asteroid.ai/) and see browser automation in action!", "url": "https://www.ycombinator.com/launches/MtY-asteroid-browser-agents-with-human-oversight", "votes": 19, "created_at": "2025-02-25T16:20:12.618Z" } ] }, { "id": 30434, "name": "Zeon Systems", "slug": "zeon-systems", "former_names": [ "Zeon Technologies" ], "small_logo_thumb_url": "https://bookface-images.s3.amazonaws.com/small_logos/95786f7359527c0b6051a250e5f3cb00d43b578b.png", "website": "https://www.zeonsystems.ai/", "all_locations": "San Francisco, CA, USA", "long_description": "We’re building AI-powered systems that automate manual work in scientific labs using robotics.\r\n\r\nScientists simply type their experiment in plain English. Zeon translates it into code and runs it on robotic arms (no coding required).\r\n\r\nBy controlling robots with natural language, we’re enabling automation for every lab, every scientist, and every experiment.\r\n\r\nZeon is lab-aware. It detects equipment, adapts to protocols, and moves autonomously between instruments to execute full workflows.\r\n\r\nWe’re already working with labs at Stanford and UCSF to accelerate discovery - from running overnight nanoparticle fluorescence measurements to safely disposing of hundreds of bacterial samples. \r\n\r\nZeon reclaims time for what matters most: the science.", "one_liner": "AI-powered robotics for scientific lab automation.", "team_size": 5, "industry": "Industrials", "subindustry": "Industrials -> Manufacturing and Robotics", "launched_at": 1745428848, "tags": [ "Robotic Process Automation", "Robotics", "Automation", "AI" ], "tags_highlighted": [], "top_company": false, "isHiring": false, "nonprofit": false, "batch": "Spring 2025", "status": "Active", "industries": [ "Industrials", "Manufacturing and Robotics" ], "regions": [ "United States of America", "America / Canada" ], "stage": "Early", "app_video_public": false, "demo_day_video_public": false, "app_answers": null, "question_answers": null, "url": "https://www.ycombinator.com/companies/zeon-systems", "api": "https://devasheeshg.github.io/yc-api/batches/spring-2025/zeon-systems.json", "year_founded": 2025, "city": "San Francisco", "country": "US", "linkedin_url": "https://www.linkedin.com/company/zeon-systems/", "x_url": "https://x.com/zeonsystems", "fb_url": null, "cb_url": null, "github_url": null, "logo_url": "https://bookface-images.s3.amazonaws.com/small_logos/95786f7359527c0b6051a250e5f3cb00d43b578b.png", "app_video_url": null, "dday_video_url": null, "company_photos": [], "primary_partner": { "name": "David Lieb", "url": "https://www.ycombinator.com/people/david-lieb" }, "founders": [ { "user_id": 379901, "full_name": "Brontë Kolar", "title": "Founder/CEO", "founder_bio": "cofounder & CEO @ Zeon Systems (YC X25)\n-\naccelerating science with robotics + AI.\n-\nbuilt electronic systems for electric aircraft and high-density batteries before pivoting to computational biology research at UCSF and UPenn. early engineer at LatchBio.", "is_active": true, "linkedin_url": "https://www.linkedin.com/in/bronte-kolar", "x_url": "", "avatar_thumb_url": "https://bookface-images.s3.us-west-2.amazonaws.com/avatars/ef08976f1f1aa308a330e07e05531de19928bf84.jpg" }, { "user_id": 2502032, "full_name": "Tahir D'Mello", "title": "Founder/CTO", "founder_bio": "cofounder & CTO @ Zeon Systems (YC X25)\naccelerating science with robotics and AI.\n\n----------------\n\nbeen building software and ML for science labs basically all my adult life, doing computational research at IIT, Yale and working in biotech and pharma across India and the US.", "is_active": true, "linkedin_url": "https://www.linkedin.com/in/tahirdme/", "x_url": "https://x.com/TahirMello", "avatar_thumb_url": "https://bookface-images.s3.us-west-2.amazonaws.com/avatars/437000c0ac09fa8258a3cafc8d88e43d147e15be.jpg" } ], "jobs": [], "news": [], "launches": [ { "id": 89951, "title": "Zeon Systems: AI-powered robotics for lab automation 🧪", "tagline": "We turn natural language into real-time lab automation.", "body": "Hey everyone 👋\n\nWe’re [Brontë](https://www.linkedin.com/in/bronte-kolar/) and [Tahir](https://www.linkedin.com/in/tahirdme/) – co-founders of [Zeon Systems](https://www.zeonsystems.ai/). We’re on a mission to make science move at the speed of thought.\n\n# **TL;DR**\n\n**We turn natural language into lab automation.** Scientists type their experiment in English. Our software programs robots to execute it.\\\n\n\n# **Problem**\n\nThe biggest scientific breakthroughs -- _think Selman Waksman mining microbes for antibiotics or Akira Yoshino testing early lithium-ion batteries_ _at Sony_ -- started in labs through painful, manual work. **Decades later, discovery still relies on pipetting, weighing, and prepping by hand.**\n\nIt’s inefficient, exhausting, and slows down discovery. Automation solutions exist, but are often **$$$** or built for **specific use cases**. \n\nWe want to change that.\n\n# **Our Take**\n\nWhat if you could run an experiment by just typing it out? \n\nThat’s what we’re building for: **automation that’s usable by everyday scientists.**\n\n# **The Zeon System 🦾**\n\nWe sell a hardware + software system that runs lab experiments using off-the-shelf robotics. **Labs receive a ready-to-run automation setup** that includes:\n\n* Our AI-powered software\n* 2 off-the-shelf robotics arms with mounted depth cameras\n\n_How it works:_\n\n* Scientists describe protocols in plain English\n* Zeon interprets them, detects lab tools using computer vision, and plans the execution\n* The system generates and executes robotic code in real-time, adapting to the specific lab environment\n\n![uploaded image](/media/?type=post&id=89951&key=user_uploads/379901/f18d8421-74dc-4213-8f17-a91419504529)\n\n# **What’s Possible**\n\nWe are already piloting with labs at UCSF and Stanford to:\n\n* Run fluorescence screens for nanoparticles overnight\n* Automate multichannel pipetting for clinical assays\n* Dispose of biohazardous waste while researchers sleep\n\n# **Big Picture**\n\nEventually, we want to **enable labs that run themselves**. Think self-driving labs on Mars or in the middle of the ocean doing science while humans dream. 🪐 🧪🧬\n\n# **The Team**\n\n[Brontë](https://www.linkedin.com/in/bronte-kolar/) built electronic systems for electric aircraft and high-density battery packs before pivoting to computational biology research at UCSF, UC Berkeley, and Erasmus Medical Center. \n\n[Tahir](https://www.linkedin.com/in/tahirdme/) has been building software and ML for science labs basically all his adult life, doing computational research at IIT, Yale and working in biotech and pharma across India and the US.\n\n# \n\n# **Our Ask**\n\nWe are looking for **pilot partners** interested in deploying our system in their labs! \n\n* **Know a scientific lab or company that should work with us?** → Email us at [founders@zeonsystems.ai](mailto:founders@zeonsystems.ai) \n* **Resonate with our message?** → Share this post (or our [promo video](https://www.youtube.com/watch?v=x-jid99QPFA))\n\n—\n\n[_Let’s make science move faster._](https://www.zeonsystems.ai/)\n\n\\-Brontë + Tahir\n\n![uploaded image](/media/?type=post&id=89951&key=user_uploads/379901/3896ab4e-84de-4bfd-ad2e-7fd6f0462be3)\n\n", "url": "https://www.ycombinator.com/launches/NOp-zeon-systems-ai-powered-robotics-for-lab-automation", "votes": 65, "created_at": "2025-04-29T23:00:50.835Z" } ] }, { "id": 30647, "name": "Cyberdesk", "slug": "cyberdesk", "former_names": [], "small_logo_thumb_url": "https://bookface-images.s3.amazonaws.com/small_logos/b6deea0aa4e96211ac35f75ade0705df19758a8e.png", "website": "https://www.cyberdesk.io", "all_locations": "San Francisco, CA, USA", "long_description": "Cyberdesk is the first self learning computer use agent for developers to automate desktop workflows. \r\n\r\nFor example, one of our customers is building an AI agent for hospital patient intake, and they’re using us to automate an entire team of people manually clicking and typing into a desktop app to book appointments. \r\n\r\nWe launched mid July, and our agent is already operating in production environments, creating appointments in electronic health record apps and processing files in legacy ERPs.\r\n\r\nWe’ve built the first computer use agent that memorizes tasks and repeats them quickly while still adapting to unexpected changes on the screen. A good analogy is to how humans use computers: once we get used to an app, we use it much faster, while still adapting to random popups and crashes. This new paradigm for computer use will be needed by every company automating human workflows on computers.\r\n\r\nMahmoud previously built robotic process automation solutions impacting 20K+ employees at a Fortune 100 company. He also built AI & ML solutions at Baylor College of Medicine Neuroscience & Texas Children's Hospital + co-published a paper at the Rice Networks Group. \r\n\r\nAlan is ex-NASA and previously dropped out to be the first founding engineer at a PearX startup. He is a second time founder, having previously built and scaled a consumer AI app to thousands of paying users.\r\n\r\nComputer-use agents will automate the $6T dollars spent on enterprise knowledge work and Cyberdesk will be the platform to power this.\r\n\r\nWe’d love to talk with you!\r\nhttps://cal.com/mahmoud-al-madi-klrs5s/30min", "one_liner": "Self learning computer use agent for developers", "team_size": 2, "industry": "B2B", "subindustry": "B2B -> Engineering, Product and Design", "launched_at": 1749659433, "tags": [ "Developer Tools", "Robotic Process Automation", "Workflow Automation", "Enterprise Software", "AI" ], "tags_highlighted": [], "top_company": false, "isHiring": false, "nonprofit": false, "batch": "Summer 2025", "status": "Active", "industries": [ "B2B", "Engineering, Product and Design" ], "regions": [ "United States of America", "America / Canada" ], "stage": "Early", "app_video_public": false, "demo_day_video_public": false, "app_answers": null, "question_answers": [ { "question": "How did your company get started? (i.e., How did the founders meet? How did you come up with the idea? How did you decide to be a founder?)", "answer": "We met 2 years ago, connecting over our shared experiences at Rice, and our love of building things. We both ended up living at the same Hacker House in Menlo Park, called Incepto! We started Cyberdesk with the realization that it’s really hard to build a great computer use agent, especially when it comes to reliability, speed, and cost. We both lived in a 9×9 Airbnb room with a tiny bed and a single desk (we call it the Cyberdesk lol) and got Cyberdesk off the ground, building the MVP in a 6 days and getting a paid customer the next day through a cold email. \\\r\n\\\r\nCyberdesk was built to solve all three of the core problems with off the shelf computer use models: speed, reliability, and cost. We’re excited to deploy this architecture to the world and become the best developer framework for computer use." }, { "question": "How did you decide to apply to Y Combinator? What was your experience applying, going through the batch, and fundraising at demo day?", "answer": "From the moment we started working together, we knew we wanted to go through YC together. It was a little over a month before the deadline for Summer 2025 came out when we started working together, so we got to work! We launched, got paid customers, and built out our infrastructure fast. A month later, we interviewed in person and got the call from Jon Xu the same night that we got in! It was one of the happiest days of our lives." }, { "question": "What's the history of your company from getting started until the present day? What were the big inflection points?", "answer": "We started working together in late March (2025), with a shared goal of solving deep developer problems around computer use. We started off with an idea to help developers spin up virtual desktops for computer use agents in a single line of code. We shipped an MVP in 5 days and got a paid customer the next day through a cold email. For the next month, we kept shipping and looking for customers, and applied to YC with this idea. We found the product had interest from hobbyist developers, but for companies, there was much less pull because companies often already have VM infrastructure they are happy with. \\\r\n\\\r\nSo we made the call to pivot. We talked to over 100 developers in the computer use and robotic process automation space, and heard the same thing over and over: automating Windows desktops is an important yet extremely tedious process. You create a script once, then it breaks every time something changes in the app or a popup happens. Computer use agents were meant to solve that pain but even they were slow, costly, and non deterministic. That’s how we came up with the idea for what Cyberdesk is today - a layer on top of computer use agents that helps it memorize tasks it’s done before and repeat it consistently, 3 - 5x faster, and cheaply, while still falling back to computer use during popups and anomalies. We built an MVP in a week, and got our first pilot a week later. \\\r\n\\\r\nSince then we’ve been rapidly building out the product and closing more deals with startups and companies in healthcare, construction, accounting, and more, all of them using our agent to automate manual computer work they’ve struggled to automate before. " }, { "question": "What is the core problem you are solving? Why is this a big problem? What made you decide to work on it?", "answer": "For a developer team to automate legacy desktop apps with poor APIs, they’ve often resorted to building brittle automation scripts (either in Python or using a legacy RPA tool like UiPath) or hiring unreliable remote workers to click on remote desktops. As they need to support more and more software and versions of that software, the development and maintenance times exponentially increases. \\\r\n\\\r\nEvery company in the world has repetitive manual desktop work that is bogging them down. On top of that, companies building AI agents are increasingly having to interface with desktop apps to perform full human workflows. This problem is pervasive and we’re excited to solve it." }, { "question": "What is your long-term vision? If you truly succeed, what will be different about the world?", "answer": "Traditional robotic process automation is a $6 billion industry, yet it only services a tiny fraction of desktop tasks that can now be automated by our agent. In the long term, our agent will automate every manual computer task." } ], "url": "https://www.ycombinator.com/companies/cyberdesk", "api": "https://devasheeshg.github.io/yc-api/batches/summer-2025/cyberdesk.json", "year_founded": 2025, "city": "San Francisco", "country": "US", "linkedin_url": "https://www.linkedin.com/company/cyberdeskhq", "x_url": "https://x.com/CyberdeskHQ", "fb_url": null, "cb_url": "https://www.crunchbase.com/organization/cyberdesk-9e2a", "github_url": "https://github.com/cyberdesk-hq/cyberdesk", "logo_url": "https://bookface-images.s3.amazonaws.com/small_logos/b6deea0aa4e96211ac35f75ade0705df19758a8e.png", "app_video_url": null, "dday_video_url": null, "company_photos": [ "https://bookface-images.s3.us-west-2.amazonaws.com/attachments/f97e92ea8e68598a95edd550d0062da113b863a0.png", "https://bookface-images.s3.us-west-2.amazonaws.com/attachments/afa862d3c6811c2f8fa968b87c2ce6d4508a98df.png" ], "primary_partner": { "name": "Jon Xu", "url": "https://www.ycombinator.com/people/jon-xu" }, "founders": [ { "user_id": 1707002, "full_name": "Mahmoud Al-Madi", "title": "Founder/CEO", "founder_bio": "Mahmoud graduated from Rice University with a degree in Electrical Engineering where he built AI & ML solutions at Baylor College of Medicine Neuroscience & Texas Children's Hospital + co-published a paper at the Rice Networks Group.\n\nAfter Rice, Mahmoud built RPA solutions impacting 20K+ employees at a Fortune 100 company, which inspired him to cofound Cyberdesk to build a computer use agent developers can use to automate desktop workflows.", "is_active": true, "linkedin_url": "https://www.linkedin.com/in/mahmoudalmadi/", "x_url": "https://x.com/mahmoudalmadi_", "avatar_thumb_url": "https://bookface-images.s3.us-west-2.amazonaws.com/avatars/61b7bae3551c4559b2db78736f13d61532e2c0b0.jpg" } ], "jobs": [], "news": [], "launches": [ { "id": 92540, "title": "Cyberdesk - The computer use agent for developers to automate legacy Windows apps", "tagline": "Automate legacy apps like EHRs, ERPs, and more with just a prompt and an API call to our agent", "body": "Hey everyone! We’re Mahmoud and Alan, co-founders of Cyberdesk.\n\n**🏆 TL;DR:**\n\nWe’re building a computer use agent for developers to automate legacy Windows apps like EHRs, ERPs, and more. Just write your task in plain English, and call our API to trigger our agent. Our agent is designed to follow memorized steps reliably, only falling back to computer use models during expected popups.\n\nhttps://www.youtube.com/watch?v=yCDGsiCNb3E\n\n**❌ The Problem**\n\nDevelopers in industries like healthcare, supply chain, and beyond often work with clients who have old, legacy Windows software. Today, to work with these clients, they probably:\n\n* Hire unreliable offshore teams to manually click through remote desktops\n* Write fragile automation scripts that break on popups and need constant fixes\n* Skip the integration entirely—leaving valuable customers behind\n\n**✨ Our Solution**\n\nJust write your task in plain English, and our computer use agent will run it reliably, cheap, and fast, on any Windows PC (cloud or on prem).\n\nExample: entering patient data into OpenDental, a legacy dental EHR (real use case from one of our customers)\n\n1. **Write the task** (”Log in to Opendental, click Patients, find this dropdown, etc”), then just call our API to trigger the workflow\n2. **Our agent learns the task.** Every time it repeats the task, it does it 100% deterministically, upwards of 3x faster, and almost zero cost\n3. **If something unexpected happens** (like a popup), our system falls back to the computer use agent, and memorizes that trajectory too\n\n![uploaded image](/media/?type=post&id=92540&key=user_uploads/1707002/d09c26db-4734-4c16-8324-e5e0657e6512)\n\n**💯 The Team**\n\nMahmoud (CEO) previously scaled robotic process automation solutions to 20k employees at a Fortune 100. Alan (CTO) is ex-NASA and built a computer use agent at his previous startup.\n\n**Book a demo with us, or intro us to AI companies in regulated industries who need to integrate into legacy desktop apps!**\n\n* Book demo: https://cal.com/mahmoud-al-madi-klrs5s/30min\n* Our website: https://cyberdesk.io\n* Email us: mahmoud \\[at\\] cyberdesk \\[dot\\] io\n\n![uploaded image](/media/?type=post&id=92540&key=user_uploads/1707002/ca98113f-44e9-44ce-ab9b-20e5ea40354a)\n\n", "url": "https://www.ycombinator.com/launches/O4a-cyberdesk-the-computer-use-agent-for-developers-to-automate-legacy-windows-apps", "votes": 34, "created_at": "2025-07-29T15:47:24.669Z" } ] }, { "id": 30991, "name": "Forge Robotics", "slug": "forge-robotics", "former_names": [], "small_logo_thumb_url": "https://bookface-images.s3.amazonaws.com/small_logos/fc718543c8da7c7d1c6d5308ecd4cde3ca18269c.png", "website": "https://forge-robotics.com", "all_locations": "San Francisco, CA, USA", "long_description": "Forge Robotics is changing the way that metal parts are fabricated. We are enabling a future where our intelligent robotic system completes all steps needed to take raw metal stock and transform it into fully welded parts. \r\n\r\nBeginning with welding, we are building a robot-mounted vision system and AI feature detection that creates a real-time 3D map and tells the arm exactly where to go at every instant.", "one_liner": "A robotics intelligence system for metal fabrication", "team_size": 1, "industry": "Industrials", "subindustry": "Industrials -> Manufacturing and Robotics", "launched_at": 1762659715, "tags": [ "Robotic Process Automation", "Robotics", "Industrial" ], "tags_highlighted": [], "top_company": false, "isHiring": false, "nonprofit": false, "batch": "Fall 2025", "status": "Active", "industries": [ "Industrials", "Manufacturing and Robotics" ], "regions": [ "United States of America", "America / Canada" ], "stage": "Early", "app_video_public": false, "demo_day_video_public": false, "app_answers": null, "question_answers": null, "url": "https://www.ycombinator.com/companies/forge-robotics", "api": "https://devasheeshg.github.io/yc-api/batches/fall-2025/forge-robotics.json", "year_founded": 2025, "city": "San Francisco", "country": "US", "linkedin_url": "https://www.linkedin.com/company/forge-robotics", "x_url": "", "fb_url": null, "cb_url": null, "github_url": null, "logo_url": "https://bookface-images.s3.amazonaws.com/small_logos/fc718543c8da7c7d1c6d5308ecd4cde3ca18269c.png", "app_video_url": null, "dday_video_url": null, "company_photos": [], "primary_partner": { "name": "Nicolas Dessaigne", "url": "https://www.ycombinator.com/people/nicolas-dessaigne" }, "founders": [ { "user_id": 2972389, "full_name": "Robert Cormican", "title": "Founder/CTO", "founder_bio": "CEO and Founder of Forge Robotics. With a deep background in Mechatronics, Image Quality and Computer Vision and Silicon Development/Semiconductor Manufacturing", "is_active": true, "linkedin_url": "https://www.linkedin.com/in/robert-cormican-a4405990/", "x_url": "", "avatar_thumb_url": "https://bookface-images.s3.us-west-2.amazonaws.com/avatars/60f5c6a0a8856605cc957c1a4ba6d20ea552717d.jpg" } ], "jobs": [], "news": [], "launches": [ { "id": 95362, "title": "Forge Robotics - The Future of Manufacturing", "tagline": "Forge Robotics gives manufacturing companies the ability to make parts fully autonomously", "body": "Hey everyone, we’re [Eoin ](https://www.linkedin.com/in/eoincobbe)and [Robert](https://www.linkedin.com/in/robert-cormican-a4405990), co-founders of Forge Robotics.\n\nWe are completely changing the way that metal fabrication is done. Companies can no longer afford to deal with massive labour shortages and rising costs of outdated processes. Forge is on the journey to enable companies to take in raw stock metal and output completed metal parts without human intervention.\n\n# TL;DR\n\nForge is building a system that turns dumb, human-programmed industrial welding robot arms into autonomous members of the workforce.\n\nFrom here, we will deliver autonomous factory units that handle all processes in metal fabrication, from handling cut parts and positioning them for assembly, to welding and cleaning up final assemblies.\n\nTo get involved, contact us @ [eoin@forge-robotics.com](https://eoin@forge-robotics.com)\n\nhttps://www.youtube.com/watch?v=5vxp75bUlIY&t=38s\n\n# 🏭 The Problem\n\nManufacturing is an industry that has been all but left behind for the past 20 years compared to the advances in software development, self-driving vehicles, and all other fields that AI has revolutionised.\n\n**Labour shortages**\n\nManufacturing of defence equipment, construction machines, and automotive make up the backbone of much of the American economy. In the U.S, there are 400,000 unfilled welding positions, and that is expected to rise to 600,000 by 2030.\n\n**Push to reshore**\n\nOEMs and Tier 1 manufacturers are pushing to reshore their manufacturing, but traditional methods of operation mean that they are failing to operate profitably when competing with manufacturing in offshore economies.\n\n**Failure of current automation**\n\nCurrent industrial robots can only achieve a very narrow set of well-defined manufacturing tasks because of the way robots are programmed. Traditionally programmed welding robots end up with around 10% of parts needing rework due to the delta in geometry from part to part.\n\n![uploaded image](/media/?type=post&id=95362&key=user_uploads/2795202/eb13105f-47fa-4312-b8c7-185ecac7e04a)\n\n# 🦾 Our solution\n\nForge provides a vision system and robotic intelligence that creates a sub-mm map, identifies all features and weld seams, and autonomously controls the welding robot arms.\n\nMetal fabrication companies will now:\n\n* No longer have to **manually program** robotic arms\n* Achieve **99% first pass line yield**, up from the current 90%\n* Use robotic welding arms in many previously infeasible situations\n\n![uploaded image](/media/?type=post&id=95362&key=user_uploads/2795202/82ef4c04-835c-4a20-a6b6-85698177d54b)\n\n# ♾️ Fully autonomous production\n\nUsing the manufacturing intelligence system that we are working on for factories today, we will be delivering fully autonomous modular factory units that handle all of the metal manufacturing processes to take raw steel parts to finished welded assemblies. Saving companies **Tens of millions** per year by changing the way they manufacture.\n\n# 🤝 Ask\n\nCan you introduce us to companies in the metal manufacturing industry? We are currently taking on as many pilot customers as we can!\n\nIf this is a future that excites you and you want to find out how you can be part of it 👉Contact Eoin @ [eoin@forge-robotics.com](https://eoin@forge-robotics.com)\n\nFind out more @ [www.forge-robotics.com](https://forge-robotics.com/)", "url": "https://www.ycombinator.com/launches/Oo6-forge-robotics-the-future-of-manufacturing", "votes": 285, "created_at": "2025-11-10T15:00:31.012Z" } ] }, { "id": 31145, "name": "RamAIn", "slug": "ramain", "former_names": [ "Ramain" ], "small_logo_thumb_url": "https://bookface-images.s3.amazonaws.com/small_logos/1a10a6d3695c38f916de546238a447f1633ea982.png", "website": "https://ramain.ai", "all_locations": "San Francisco, CA, USA", "long_description": "RamAIn Employees join your team just like a human. Teach them to work across systems just like a new hire and watch them work autonomously to do the work 24/7. ", "one_liner": "RamAIn - AI Employees with a computer to perform business tasks just…", "team_size": 7, "industry": "B2B", "subindustry": "B2B", "launched_at": 1767663208, "tags": [ "Robotic Process Automation", "Automation", "AI" ], "tags_highlighted": [], "top_company": false, "isHiring": true, "nonprofit": false, "batch": "Winter 2026", "status": "Active", "industries": [ "B2B" ], "regions": [ "United States of America", "America / Canada", "Remote", "Partly Remote" ], "stage": "Early", "app_video_public": false, "demo_day_video_public": false, "app_answers": null, "question_answers": null, "url": "https://www.ycombinator.com/companies/ramain", "api": "https://devasheeshg.github.io/yc-api/batches/winter-2026/ramain.json", "year_founded": 2025, "city": "San Francisco", "country": "US", "linkedin_url": "https://www.linkedin.com/company/ramain", "x_url": "https://x.com/ramainai", "fb_url": null, "cb_url": "https://www.crunchbase.com/organization/ramain", "github_url": null, "logo_url": "https://bookface-images.s3.amazonaws.com/small_logos/1a10a6d3695c38f916de546238a447f1633ea982.png", "app_video_url": null, "dday_video_url": null, "company_photos": [], "primary_partner": { "name": "Tyler Bosmeny", "url": "https://www.ycombinator.com/people/tyler-bosmeny" }, "founders": [ { "user_id": 2267746, "full_name": "Shourya Vir Jain", "title": "Founder/CEO", "founder_bio": "Building the world's fastest computer-use agents for enterprise work 🍜\n- EE @ IIT Delhi\n- FIDE Chess Rating 2118, top 20 in the world under 14\n- ex-McKinsey & Company\n", "is_active": true, "linkedin_url": "https://www.linkedin.com/in/shouryavirjain/", "x_url": "https://x.com/SveeVJ", "avatar_thumb_url": "https://bookface-images.s3.us-west-2.amazonaws.com/avatars/2b48edd363a4d2807e1d93d6d846d416b79e279c.jpg" }, { "user_id": 3198910, "full_name": "Vansh Ramani", "title": "Founder/CTO", "founder_bio": "Building the world's fastest computer-use agents for enterprise work 🍜\n- CS @ IIT Delhi\n- AI researcher at CMU working on scalable ML and representation learning\n- Published research papers at ICLR, ICML and ACS including 'Panaroma' - world's fastest vector search algorithm - later merged into Meta FAISS", "is_active": true, "linkedin_url": "https://www.linkedin.com/in/ramanivansh/", "x_url": "https://x.com/vansh_ramani", "avatar_thumb_url": "https://bookface-images.s3.us-west-2.amazonaws.com/avatars/9b1b32f82ebad1410fd58a4ab062931a79c5c7dc.jpg" } ], "jobs": [ { "id": 92604, "title": "Founding AI/ML Research Engineer", "url": "https://www.ycombinator.com/companies/ramain/jobs/jezgwo5-founding-ai-ml-research-engineer", "location": "San Francisco, CA, US", "type": "Full-time", "role": "Engineering", "role_type": "Full stack", "salary_range": "$150K - $350K", "equity_range": "0.30% - 1.20%", "experience": "1+ years", "visa": "US citizen/visa only", "skills": [ "Python", "Machine learning" ] }, { "id": 95468, "title": "Founding GTM Engineer", "url": "https://www.ycombinator.com/companies/ramain/jobs/hqvmyKN-founding-gtm-engineer", "location": "San Francisco, CA, US", "type": "Full-time", "role": "Sales", "role_type": null, "salary_range": "$120K - $180K", "equity_range": "0.30% - 1.20%", "experience": "1+ years", "visa": "US citizen/visa only", "skills": [] } ], "news": [], "launches": [ { "id": 97571, "title": "RamAIn: Teach AI to use your legacy systems, desktop applications and web portals just like a human", "tagline": "With AI, we handle complex data entry and extraction operations on which RPA breaks. 10x faster than computer use agents from Google, OpenAI and Anthropic.", "body": "# \\*\\*\\*\\*\n\n**tl;dr: [RamAIn](https://www.ramain.ai/)** _intelligently_ _automates complex and repetitive work across browser and desktop applications. It doesn’t matter if you have API access or not - the general principle is: **if you can see it, RamAIn can do it.**_\n\n[RamAIn Launch Video](https://www.youtube.com/watch?v=tnVTskMw3mk)\n\nEnterprise teams are stuck **manually reading, copy-pasting and tab-switching** for hours every week. RamAIn kills the busywork so your team can get back to the real work.\n\nRamAIn agents simulate mouse and keyboard clicks to use your team’s computers just like a human to move data between your **legacy systems, desktop applications and web portals.** It dominates complex data entry and extraction tasks through UI-interaction on which traditional RPA breaks.\n\nRamAIn is **self-healing** and great at handling unexpected pop-ups or UI changes by using AI to solve for deviations it hasn’t seen before. It takes just a **few days to train and deploy** across your team’s applications and maintains a full audit-trail of decisions.\n\nUnlike current computer-use AI agents that have to think about every action while executing, **RamAIn learns UI-policies and interface structures** from your key interfaces that allows it to make decisions 10x faster while being super reliable.\n\n![uploaded image](/media/?type=post&id=97571&key=user_uploads/2267746/6806e0de-aca0-4f07-a2c4-5a8c2a53408f)\n\n_What RamAIn sees after it has learnt how to use Outlook_\n\nRamAIn is super-safe. It comes out of the box with human-in-the-loop support, giving your team real-time control over every workflow. It’s designed to augment your team, boosting productivity without replacing human expertise.\n\nThrough a simple UI overlay, RamAIn can communicate with your team in real-time, asking clarifying questions whenever context is missing or expert judgment is required.\n\n![uploaded image](/media/?type=post&id=97571&key=user_uploads/2267746/0f41ab6f-7efb-4316-a522-2cafe22191a2)\n\n_RamAIn requesting human-input in assisted mode_\n\nRamAIn unifies automations into a single agent by integrating with APIs when required and using UI-interaction where APIs aren’t available to ensure complete task coverage across both internal and 3rd-party systems.\n\n![uploaded image](/media/?type=post&id=97571&key=user_uploads/2267746/fc2923e3-da81-4140-a931-9f2964350517)\n\n_RamAIn filling up a dynamic 3rd-party form in real-time_\n\n**We’ve learnt through our customers that RamAIn is perfect for -**\n\n**Procurement Teams:** \\\n_(ERP + Supplier Portals)_\\\nVendor onboarding, PO creation, invoice matching, supplier data entry.\n\n**Insurance Brokers:** \\\n_(Carrier Portals + Brokering Platforms)_\\\nSubmit client intake across carrier portals, extract quotes, compare premiums, update internal systems.\n\n**Healthcare Providers:** \\\n_(EHRs + Lab & Payer Portals)_\\\nTransfer patient data between intake forms, manage prior authorizations, coordinate labs and payers.\n\n**Revenue Cycle Management Teams:** \\\n_(RCM Platforms + Payer Portals)_\\\nEligibility checks, claim submissions, denial follow-ups, payment posting, EOB reconciliation.\n\n**Finance & Operations Teams:** \\\n_(ERP + Legacy Finance Systems)_\\\nAP/AR automation, reconciliations, reporting, inventory updates.\n\n**Retail & Store Operations:** \\\n_(POS + Accounting Systems)_\\\nSales reconciliation, pricing updates, transaction reporting.\n\n**Pharmacy Operations:**\\\n_(Pharmacy Management Systems + Distributor Portals)_\\\nPrescription processing, insurance claim submission, inventory reconciliation.\n\n**Logistics & Transportation Teams:** \\\n_(TMS + Carrier Portals)_\\\nLoad booking, carrier updates, shipment tracking, document uploads, freight invoice reconciliation.\n\n**Is your team also drowning in manual labour?**\n\nSign up to be a design partner today —> [www.ramain.ai/automate](https://www.ramain.ai/automate)\n\n![uploaded image](/media/?type=post&id=97571&key=user_uploads/2267746/16773a2a-f59e-4e17-af88-e5d0c65c90a2)\n\n", "url": "https://www.ycombinator.com/launches/PNj-ramain-teach-ai-to-use-your-legacy-systems-desktop-applications-and-web-portals-just-like-a-human", "votes": 150, "created_at": "2026-02-06T03:22:35.642Z" } ] }, { "id": 31428, "name": "Ontora", "slug": "ontora", "former_names": [], "small_logo_thumb_url": "https://bookface-images.s3.amazonaws.com/small_logos/51ba58b94e0a45c3a3cc270067fe72dcf1f9027e.png", "website": "https://ontora.com", "all_locations": "Berlin, Berlin, Germany", "long_description": "Ontora is an AI Agent that interviews every employee to map how work actually gets done and identify bottlenecks. Consultants spend months doing this manually; Ontora does it in hours.\r\n\r\n", "one_liner": "AI agents that interviews every employee to hand context to AI tools. ", "team_size": 3, "industry": "B2B", "subindustry": "B2B -> Operations", "launched_at": 1773487694, "tags": [ "Robotic Process Automation", "Enterprise", "AI" ], "tags_highlighted": [], "top_company": false, "isHiring": false, "nonprofit": false, "batch": "Spring 2026", "status": "Active", "industries": [ "B2B", "Operations" ], "regions": [ "Germany", "Europe" ], "stage": "Early", "app_video_public": false, "demo_day_video_public": false, "app_answers": null, "question_answers": [ { "question": "How did your company get started? (i.e., How did the founders meet? How did you come up with the idea? How did you decide to be a founder?)", "answer": "Max led process optimization at Porsche. David built data pipelines in finance. Leon built AI automations at consultancies. They all noticed critical knowledge is trapped inside people's heads and never reaches the people who need it.\r\n\r\nThe team then felt inspired to work on the problem after a tour to Stockholms fastest scaling startups organized by Max, where all three decided to quit their jobs and start a company. " }, { "question": "What is the core problem you are solving? Why is this a big problem? What made you decide to work on it?", "answer": "Every company already has the answers to most of the questions leadership is asking.\r\n\r\nThe problem: those answers are scattered across hundreds of people, and there's never been a practical way to ask every single employee.\r\n\r\nSo leaders make crucial decisions with partial information or spend months and millions on external consultants to retrieve what the org already knows.\r\n\r\n**What we're building**\r\n\r\nOntora is an AI agent that has a real conversation with every employee in your company within hours.\r\n\r\nIt learns, follows up, and surfaces what people actually do and where they get stuck.\r\n\r\nWe deliver two things: a synthesized answer to the specific question leadership couldn't crack in their last business review (themed insights, process maps, automation roadmap with ROI), and a continuously refreshed operational context that humans and AI agents can both reason over." }, { "question": "What is your long-term vision? If you truly succeed, what will be different about the world?", "answer": "Answers in hours instead of months. We imagine truly intelligence native companies that unlock the last proprietary data source left in the enterprise: operational knowledge living in your employees heads.\r\n\r\nEnabling cost reduction, post-merger integration, and AI transformation… within hours and at a fraction of the cost of traditional consulting." } ], "url": "https://www.ycombinator.com/companies/ontora", "api": "https://devasheeshg.github.io/yc-api/batches/spring-2026/ontora.json", "year_founded": 2026, "city": "Berlin", "country": "DE", "linkedin_url": "https://www.linkedin.com/company/ontora/", "x_url": "https://x.com/ontoratech", "fb_url": null, "cb_url": "https://www.crunchbase.com/organization/ontora", "github_url": null, "logo_url": "https://bookface-images.s3.amazonaws.com/small_logos/51ba58b94e0a45c3a3cc270067fe72dcf1f9027e.png", "app_video_url": null, "dday_video_url": null, "company_photos": [ "https://bookface-images.s3.us-west-2.amazonaws.com/attachments/77783f688ca955fad6bf02d0bcff0d2288011691.png", "https://bookface-images.s3.us-west-2.amazonaws.com/attachments/2750bcd85d7a83625c723e8099b143de9e784edb.png" ], "primary_partner": { "name": "Tom Blomfield", "url": "https://www.ycombinator.com/people/tom-blomfield" }, "founders": [ { "user_id": 1846571, "full_name": "Leon Iwanowitsch", "title": "Founder/CEO", "founder_bio": "AI Consulting for Enterprise; 🇨🇭 University of St.Gallen (Europe's #1 Business School), Built Europe's largest Early-Stage StartUp Conference (7k participants e.g. Eric Schmidt), Helped raise 15m Solo GP Fund, Angel Investor", "is_active": true, "linkedin_url": "https://www.linkedin.com/in/leon-j-iwanowitsch/", "x_url": "https://x.com/LeonIwanowitsch", "avatar_thumb_url": "https://bookface-images.s3.us-west-2.amazonaws.com/avatars/091999d38e676fd36c0ab9735d1282401ca33cfe.jpg" }, { "user_id": 2165528, "full_name": "Maximilian Arnold", "title": "Founder/COO", "founder_bio": "1st AI Engineer @ Porsche in procurement. CS at CODE University and Stanford. Won +$80k in hackathons. Built the official honors exchange partnership to Stanford. ", "is_active": true, "linkedin_url": "https://www.linkedin.com/in/arnold-max/", "x_url": "https://x.com/maxonary", "avatar_thumb_url": "https://bookface-images.s3.us-west-2.amazonaws.com/avatars/02f8fee9ed1318020c091b986f7dca072ada839b.jpg" }, { "user_id": 1178744, "full_name": "David Korn", "title": "Founder/CTO", "founder_bio": "AI Projects @ Google, exited previous company to beglaubigt.de (YC F24), Computer Science at CODE University (founded by the largest German Unicorn founders)", "is_active": true, "linkedin_url": "https://www.linkedin.com/in/korn-david/", "x_url": "https://x.com/dav1dk0rn", "avatar_thumb_url": "https://bookface-images.s3.us-west-2.amazonaws.com/avatars/23e82b31b5c7d2f8705c94bdbec42ae0b85ac474.jpg" } ], "jobs": [], "news": [ { "title": "Y Combinator P26: Ontora KI-Agenten, Verborgenes Wissen – Ieon Iwanowitsch, Maximilian Arnold, David Korn", "url": "https://startup-insider.com/podcast/y-combinator-p26-zwei-monate-alt-ki-agenten-weltweiter-inbound-ieon-iwanowitsch-maximilian-arnold-david-korn/", "date": "Apr 27, 2026" } ], "launches": [ { "id": 102685, "title": "Ontora - The Discovery Layer for AI Transformation", "tagline": "5 enterprise customers, 150+ inbound demos (all a month after launch)", "body": "# **One month ago we launched Ontora. Here's what happened**\n\n### Our progress\n\n\\- 5 enterprise customers as design partners (many more to come ;)\\\n\\- 150+ inbound demo calls — we didn't do outbound, they came to us\\\n\\- $700K raised before officially opening our round\\\n\\- 80+ inbound VC calls already lined up for our fundraise\n\n\n\nAnd we released our product live with our first customers !\n\n**1. Why companies are banging down our door:**\n\nEvery company we talk to has the same story: they bought Claude licenses for their teams but merely use it to rewrite emails. The AI transformation stalled because nobody mapped what to actually automate.\n\nThat's the real bottleneck. Not code. Context.\n\nA 800-person European media agency flew to San Francisco to meet us after realizing they'd need 4-5 months and a team of consultants just to understand their own operations. With Ontora, they can map their entire company in days.\n\n![uploaded image](/media/?type=post&id=102685&key=user_uploads/1178744/2bd9ca28-443a-46ef-af8c-ad76dd2d6a9f)\n\n**2. What Ontora does for your company:**\n\nOntora deploys AI agents that interview every employee in your company -- not once, but continuously -- to map out how work actually gets done. We turn that into a living context layer that your AI agents can act on.\n\nWhat used to take a consulting firm 4 months and 50-100 manual interviews, Ontora does in days. And because interviews run continuously, your context layer never goes stale. Your organization keeps learning about itself.\n\n![uploaded image](/media/?type=post&id=102685&key=user_uploads/1178744/56b07749-ae1e-4244-9064-d1dd122baecc)\n\nHere's how it works:\n\n1. **Deploy** -- Ontora's AI agents reach out to employees across your org through async interviews, meeting them where they work\n2. **Map** -- Every interview extracts processes, decision rules, tools used, handoffs, and tribal knowledge into structured operational maps\n3. **Act** -- The living context layer feeds directly into your AI agents and automations so they have the real-world context they need to perform\n\n\n\n**3. Who's already using Ontora:**\n\n* **A PE firm in San Francisco** -- deploying Ontora across portfolio companies to drive operational efficiency at speed no consulting firm can match\n* **A major European media agency (800+ employees)** -- building their corporate memory to manage enterprise customers like Volkswagen and Bosch\n* **A 1,000-person European roll-up** -- mapping what their remote workforce actually does day-to-day to build a concrete AI automation roadmap\n* **A US facilities maintenance marketplace** -- extracting the proposal rules and customer preferences locked in their sales team's heads to build the context layer for their AI automations\n\n**4. We want to hear from you:**\n\n* If your mid-market company (200-2,000 employees) is planning AI transformation...\n* If your PE/VC firm wants to drive efficiency across portfolio companies...\n* If your company spent 6 figures on consultants to map processes...\n\n...we need to talk.\n\n[ontora.com/demo](https://ontora.com/demo)", "url": "https://www.ycombinator.com/launches/QiD-ontora-the-discovery-layer-for-ai-transformation", "votes": 122, "created_at": "2026-06-08T17:25:40.562Z" }, { "id": 99850, "title": "Ontora - Read your company like a book", "tagline": "We interview every employee to surface what no CEO can see", "body": "We’re [Max](https://www.linkedin.com/in/arnold-max/), [Leon](https://www.linkedin.com/in/leon-j-iwanowitsch/) and [David](https://www.linkedin.com/in/korn-david/) co-founders of [Ontora](https://ontora.com/).\\\n\\\n**TL;DR**\\\n**Ontora is an AI agent that interviews every employee in your company to map how work actually gets done.** Most operational knowledge never gets documented, it lives in people's heads. That’s crucial context inaccessible to decision-makers and AI agents.\\\n**_Ask:_** Imagine you could talk to every employee in your company within 24 hours - What would you ask? Send us your questions. We'll get you the answers: \\\n\\\nOur Launch Video: \\\n\\\n**Problem**\\\n**Poor knowledge sharing costs Fortune 500 companies $31.5 billion every year.**\n\n* **The knowledge lives in people's heads:** 80% of operational knowledge is tacit and never documented\n* 42% of what an employee produced cannot be covered by other colleagues\n* Employees lose 5.3 hours per week searching for information and filling gaps\n* A 1,000-employee company loses $2.4M per year in productivity due to these gaps\n* **AI can't read minds (yet):** 95% of enterprise GenAI pilots deliver zero measurable business impact. Agents fail because they have no context for how operations actually run.\n\n**Solution**\\\nYou pick a question your last business review meeting couldn't answer. Ontora interviews every employee and gets you the insights within 24 hours.\\\nIt’s not a survey! A real conversation that digs deeper and surfaces what nobody thought to ask about.\\\nBuild your company's operational memory to finally operate with complete clarity:\n\n* **Synthesized findings**: Themed insights with impact ratings, an executive summary, and actionable recommendations\n* **Process mapping**: Visual process maps of how work actually flows across teams\n* **Automation roadmap**: Concrete opportunities with ROI estimates, ready to act on\n* **Proprietary context for your AI tools:** Ontora turns what your people know into a proprietary data source your AI stack can finally work with.\n\n ![uploaded image](/media/?type=post&id=99850&key=user_uploads/1178744/68dcb300-dc42-4b79-aa10-9f8ec09af0f4)\n\n", "url": "https://www.ycombinator.com/launches/PyU-ontora-read-your-company-like-a-book", "votes": 144, "created_at": "2026-04-17T15:35:36.009Z" } ] }, { "id": 32154, "name": "Zomma", "slug": "zomma", "former_names": [], "small_logo_thumb_url": "https://bookface-images.s3.amazonaws.com/small_logos/019855257e473ab4938156b12e16e4a3214a2f51.png", "website": "https://zommalabs.com", "all_locations": "San Francisco, CA, USA", "long_description": "Zomma is building the AI workforce for financial services with computer-use agents that onboard like humans and scale like compute. Its agents handle KYC reviews, transaction alerts, and disputes on the same screens human analysts use, no integration.", "one_liner": "The AI BPO for financial servies, run by computer-use agents", "team_size": 5, "industry": "B2B", "subindustry": "B2B", "launched_at": 1783572526, "tags": [ "Robotic Process Automation", "Finance", "B2B", "Workflow Automation", "AI" ], "tags_highlighted": [], "top_company": false, "isHiring": false, "nonprofit": false, "batch": "Summer 2026", "status": "Active", "industries": [ "B2B" ], "regions": [ "United States of America", "America / Canada" ], "stage": "Early", "app_video_public": false, "demo_day_video_public": false, "app_answers": null, "question_answers": null, "url": "https://www.ycombinator.com/companies/zomma", "api": "https://devasheeshg.github.io/yc-api/batches/summer-2026/zomma.json", "year_founded": 2026, "city": "San Francisco", "country": "US", "linkedin_url": "https://www.linkedin.com/company/zomma", "x_url": "https://x.com/ZommaLabs", "fb_url": null, "cb_url": null, "github_url": null, "logo_url": "https://bookface-images.s3.amazonaws.com/small_logos/019855257e473ab4938156b12e16e4a3214a2f51.png", "app_video_url": null, "dday_video_url": null, "company_photos": [ "https://bookface-images.s3.us-west-2.amazonaws.com/attachments/b2b79e62b805cf93e198d9b3d4b6342664e73d7c.png" ], "primary_partner": { "name": "Brad Flora", "url": "https://www.ycombinator.com/people/brad-flora" }, "founders": [ { "user_id": 3128623, "full_name": "Jihyun Kim", "title": "Founder/CEO", "founder_bio": "Founder at Zomma (YC S26). Launched $1B Hedge Fund as the Founding Team and Head of Product. Scaled Asia's largest wealth tech (now $10B+ AUM). Second time founder. Ex-Goldman Sachs. Cornell and NYU Abu Dhabi.", "is_active": true, "linkedin_url": "https://www.linkedin.com/in/kimjihyun/", "x_url": "https://x.com/jihprobs", "avatar_thumb_url": "https://bookface-images.s3.us-west-2.amazonaws.com/avatars/67617ac3a534465e42aa00e5e395beaed235f970.jpg" }, { "user_id": 3597712, "full_name": "Varun Arumugam", "title": "Founder", "founder_bio": "Founder at Zomma (YC S26). Previously Founding ML Engineer at Bronco AI building agents for chip design verification. SWE at Apple on verification infrastructure. MS and BS in Electrical and Computer Engineering from UT Austin.", "is_active": true, "linkedin_url": "https://www.linkedin.com/in/varunarumugam/", "x_url": "", "avatar_thumb_url": "https://bookface-images.s3.us-west-2.amazonaws.com/avatars/526b6e3a7761d998cfb1049aab57c89fa1f3cd11.jpg" }, { "user_id": 3627790, "full_name": "Cashel Fitzgerald", "title": "Founder/CTO", "founder_bio": "Co-founder at Zomma. Master's in Computer Science from Cornell. Built SOTA aerial object detection systems at Modern Intelligence. Previously engineered backend systems at 8am supporting $300M+ annual revenue and $20B+ yearly transactions. Trains VLMs for computer use.", "is_active": true, "linkedin_url": "https://www.linkedin.com/in/cashelfitzgerald", "x_url": "", "avatar_thumb_url": "https://bookface-images.s3.us-west-2.amazonaws.com/avatars/69c3f4c9b45abfcd2073a8a53036f40eba0a1ce1.jpg" } ], "jobs": [], "news": [], "launches": [ { "id": 110619, "title": "Zomma: AI operations employees for finance back offices", "tagline": "Computer-use agents that work cross-border transaction alert investigations end to end: collect evidence, produce an assessment, draft an RFI.", "body": "We're building Zomma: AI operations employees for finance back offices, starting with cross-border transaction alert investigations.\n\n**TL;DR:** Zomma agents work directly inside the software your ops team already uses. No APIs, no integrations. For every cross-border alert, the agent gathers the evidence, writes the draft assessment, and drafts the RFI, end-to-end. Running live compliance-ops pilots with a US RIA and a $10B AUM wealth tech. 92% on OSWorld.\n\n\n\n## **The problem**\n\nCross-border payments trip more monitoring rules than anything else in the queue: sanctions name hits on foreign counterparties, high-risk corridors, first-time beneficiaries. Most alerts are false positives (McKinsey: over 90% in transaction monitoring), but every one still needs to be investigated. 20 to 60 minutes per case, across five or six systems, several with no API at all. And the evidence that clears them sits in foreign company registries and local-language adverse media. \n\nSo review capacity scales the only way it can: hiring. When hiring lags volume, alerts pile into backlogs that take months to clear, and backlogs are what regulators fine: $3.8B in AML penalties last year alone.\n\n## **What Zomma does**\n\nZomma agents work the queue on the same screens your analysts use: case manager, screening tool, admin panel, inbox, web search. Onboarding looks like hiring: an SOP and a login.\n\nOn a transaction monitoring alert, the agent\n\n* pulls the KYC profile and payment history\n* verifies the invoice and researches the counterparty, including registries you can only search by hand\n* writes a case-specific assessment with every claim cited to the source screen\n* drafts the RFI when documents are missing\n\nEvery step is logged and replayable: timestamped screenshots of every page checked, numbered citations, explicit no-match statements, a screen-by-screen trail an examiner can follow.\n\n## **Why start with cross-border alerts**\n\nIf an employee can do it on a screen, Zomma can learn it: the rest of the compliance queue, payment disputes, client onboarding, trade ops. Cross-border alerts come first because they have the exact shape our agents are best at: high-volume, parallelizable work that needs real reasoning but follows a defined workflow to a documented disposition. They're also the hardest cases to automate any other way: the evidence lives in registries and portals that will never have APIs. And agents scale with the queue: a volume spike means provisioning two more that afternoon; a BPO batch takes a month to train.\n\nComputer-use agents only recently got accurate enough to trust with a live queue, which is why we're building this now. Our state-of-the-art computer-use harness with our procedural and workflow memory system scores 92% on OSWorld.\n\n## **Team**\n\n* **Jihyun (CEO):** founding team at a Blackstone-backed hedge fund, scaled Asia's largest wealth tech, Goldman engineer.\n* **Cash (CTO):** backend systems behind $20B+ in yearly transactions, trains VLMs for computer use.\n* **Varun (CPO):** high-performance infra at Apple, design-verification agents at Bronco AI (YC S23).\n\nSend us \\~100 closed cross-border alerts and we'll show you what the agent would have done with them, or book a call: zommalabs.com/call.", "url": "https://www.ycombinator.com/launches/SmB-zomma-ai-operations-employees-for-finance-back-offices", "votes": 76, "created_at": "2026-08-18T16:00:00.082Z" } ] }, { "id": 32717, "name": "Neuron Industries", "slug": "neuron-industries", "former_names": [], "small_logo_thumb_url": "https://bookface-images.s3.amazonaws.com/small_logos/12fc3438894b5a05c8d3d1997a329d2c5053c9b8.png", "website": "https://neuronindustries.com", "all_locations": "Los Angeles, CA, USA", "long_description": "Neuron is crafting the world’s first purpose-built industrial controller for AI, replacing the antiquated PLCs running every machine and factory. We help customers from chemical processing to nuclear reactors compress controls projects from weeks to hours. ", "one_liner": "Industrial Controllers built for AI", "team_size": 2, "industry": "Industrials", "subindustry": "Industrials -> Manufacturing and Robotics", "launched_at": 1784835700, "tags": [ "Robotic Process Automation", "Manufacturing", "Automation", "Industrial" ], "tags_highlighted": [], "top_company": false, "isHiring": false, "nonprofit": false, "batch": "Summer 2026", "status": "Active", "industries": [ "Industrials", "Manufacturing and Robotics" ], "regions": [ "United States of America", "America / Canada" ], "stage": "Early", "app_video_public": false, "demo_day_video_public": false, "app_answers": null, "question_answers": null, "url": "https://www.ycombinator.com/companies/neuron-industries", "api": "https://devasheeshg.github.io/yc-api/batches/summer-2026/neuron-industries.json", "year_founded": 2026, "city": "Los Angeles", "country": "US", "linkedin_url": "https://www.linkedin.com/company/neuron-industries-1", "x_url": "", "fb_url": null, "cb_url": null, "github_url": null, "logo_url": "https://bookface-images.s3.amazonaws.com/small_logos/12fc3438894b5a05c8d3d1997a329d2c5053c9b8.png", "app_video_url": null, "dday_video_url": null, "company_photos": [], "primary_partner": { "name": "Diana Hu", "url": "https://www.ycombinator.com/people/diana-hu" }, "founders": [ { "user_id": 3676531, "full_name": "Dennis Ren", "title": "Founder/CEO", "founder_bio": "CEO and Co-Founder at Neuron Industries. Previously founding EE at Freeform, built its hardware tech stack from the ground up and scaled the world's most advanced metal 3D printing factory to production. Vision Pro System EE at Apple, owned eye tracking and Optic ID subsystems form factor system and board design. Ex-Tesla (Fremont) , Amazon (Go), Apple (iPad Pro, Pencil). UCSD BSEE.", "is_active": true, "linkedin_url": "https://www.linkedin.com/in/dennisren/", "x_url": "", "avatar_thumb_url": "https://bookface-images.s3.us-west-2.amazonaws.com/avatars/da38da0d216a741391a7f9784585f65188573dcf.jpg" }, { "user_id": 3732370, "full_name": "Kenneth Rhee", "title": "Founder/CTO", "founder_bio": "CTO and Co-Founder at Neuron Industries. Previously founded WorkIPO, bootstrapping it as a solo founder, building mission-critical enterprise software systems. Seller-marketplace systems at GOAT.com transacting hundreds millions of dollars, engineering at Amazon on Elastic Block Store (trillions I/O ops daily), Silk Browser for FireTV at Amazon, Dassault Systemes' Medidata on regulated clinical services. Computer Science at UC Irvine.", "is_active": true, "linkedin_url": "https://www.linkedin.com/in/kennethrhee/", "x_url": "https://x.com/kenneth_rhee", "avatar_thumb_url": "https://bookface-images.s3.us-west-2.amazonaws.com/avatars/44092f707a68aa3f8c2cff8e697c1a3f2b86ebef.jpg" } ], "jobs": [], "news": [], "launches": [ { "id": 107212, "title": "Neuron Industries: The world's first purpose built industrial controller for AI", "tagline": "Throw away your PLCs!", "body": "\n\n**TL;DR:** Neuron is building the world's first purpose built industrial controller for AI, replacing the antiquated PLCs running every machine and factory. We compress controls projects from weeks to hours, from chemical processing to nuclear reactors.\n\nWe have hardware ready to deploy today, hit us up for a pilot!\n\n**Problem**\n\nEvery piece of industrial equipment, conveyors, pumps, packaging cells, process plants, runs on PLCs. But PLCs got stuck in the 90s: it's 2026 and you're paying $5000 for a controller with 5MB of memory, programmable only with esoteric IDEs and languages frozen in 1993. Reindustrialization and data center buildups are pushing demand up while the programmer workforce shrinks. Projects still cost a fortune and take months.\n\n**Solution**\n\nWe're rebuilding the controller from first principles: state-of-the-art hardware, an AI-native stack, one ecosystem unifying PLC, HMI, SCADA, and Historian. Our agent contextualizes your requirements and hardware, then lets anyone program production applications in plain English, with test and verification guardrails and first-class data observability.\n\n**Traction**\n\nOur tech is in customers' plants today: pilots in chemical processing, electronics manufacturing, wastewater treatment, and small modular reactors.\n\n**Team**\n\nDennis Ren (CEO): founding EE at Freeform, built the hardware stack for the world's most advanced metal 3D-printing factory. Apple Vision Pro founding team, Amazon Go, Tesla Fremont.\n\nKenneth Rhee (CTO): second-time founder, bootstrapped and exited WorkIPO. Custom ERP, MES, and CRM systems; ex-Amazon and GOAT Group (YC W11).\n\n**Our ask**\n\nIf you — or someone you know — build machines or factories that run on PLCs and HMIs, we'd love to talk: [hello@neuronindustries.com](mailto:hello@neuronindustries.com)", "url": "https://www.ycombinator.com/launches/RtE-neuron-industries-the-world-s-first-purpose-built-industrial-controller-for-ai", "votes": 12, "created_at": "2026-07-24T18:02:58.034Z" } ] }, { "id": 33022, "name": "Praxis Robotics", "slug": "praxis-robotics", "former_names": [ "Praxis AI" ], "small_logo_thumb_url": "https://bookface-images.s3.amazonaws.com/small_logos/9bc18b804ad9a5d92ef1db88e90491f98e0204a7.png", "website": "https://www.praxisrobotics.io", "all_locations": "San Francisco, CA, USA", "long_description": "Praxis captures the data physical AI is bottlenecked on: egocentric video, 3D scans, and multimodal capture from inside real industrial and residential environments. Embedded within publicly listed and unicorn-scale conglomerates, we reach 60k workers across 5 continents and 150+ environment types, supplying the real-world training data frontier labs and humanoid companies can't source anywhere else.", "one_liner": "Turns real businesses into robotics data vendors", "team_size": 3, "industry": "Industrials", "subindustry": "Industrials", "launched_at": 1782948528, "tags": [ "Hard Tech", "Reinforcement Learning", "Robotic Process Automation" ], "tags_highlighted": [], "top_company": false, "isHiring": false, "nonprofit": false, "batch": "Summer 2026", "status": "Active", "industries": [ "Industrials" ], "regions": [ "United States of America", "America / Canada" ], "stage": "Early", "app_video_public": false, "demo_day_video_public": false, "app_answers": null, "question_answers": null, "url": "https://www.ycombinator.com/companies/praxis-robotics", "api": "https://devasheeshg.github.io/yc-api/batches/summer-2026/praxis-robotics.json", "year_founded": 2026, "city": "San Francisco", "country": "US", "linkedin_url": "https://www.linkedin.com/company/130274179/admin/dashboard/", "x_url": "", "fb_url": null, "cb_url": null, "github_url": null, "logo_url": "https://bookface-images.s3.amazonaws.com/small_logos/9bc18b804ad9a5d92ef1db88e90491f98e0204a7.png", "app_video_url": null, "dday_video_url": null, "company_photos": [], "primary_partner": { "name": "Diana Hu", "url": "https://www.ycombinator.com/people/diana-hu" }, "founders": [ { "user_id": 2794157, "full_name": "Rohan Seelamsetty", "title": "Founder/CEO", "founder_bio": " CEO @ Praxis. Forged across factory floors, campaign trails, and niche forums. Ex Computation and Cognition at Penn, Apollo Fellow at Oxford, and Indian Debate Team (#1 ranked team worldwide)", "is_active": true, "linkedin_url": "https://www.linkedin.com/in/rohan-s-1186b2214", "x_url": "https://x.com/ro_seelamsetty", "avatar_thumb_url": "https://bookface-images.s3.us-west-2.amazonaws.com/avatars/4b3e6e16cddb3da2384ce8f9439ca9ebc043d2bb.jpg" }, { "user_id": 3752185, "full_name": "Dev Karpe", "title": "Founder/COO", "founder_bio": "Founder at Praxis AI, previously at Wharton's Huntsman Program. Background in investment banking and supply chain ops at a F500 CEO's office. ", "is_active": true, "linkedin_url": "https://www.linkedin.com/in/dev-karpe", "x_url": "", "avatar_thumb_url": "https://bookface-images.s3.us-west-2.amazonaws.com/avatars/f4a1fc8056a887a99633492428b5db995e58a488.jpg" }, { "user_id": 3748464, "full_name": "Tommy Li", "title": "Founder/CTO", "founder_bio": "Founder at Praxis AI. Prev. tech private equity at Morgan Stanley. Industrial tech venture capital at Andreessen Horowitz. Published IEEE researcher at NASA JPL and PerceptIn Robotics. BSE CS and Finance from UPenn.", "is_active": true, "linkedin_url": "https://www.linkedin.com/in/tommyli-penn/", "x_url": "", "avatar_thumb_url": "https://bookface-images.s3.us-west-2.amazonaws.com/avatars/0e7be9454dacfd8fa65594cc6836b90cc7e5e0b4.jpg" } ], "jobs": [], "news": [], "launches": [ { "id": 109471, "title": "Praxis AI - Training Data for Robotics", "tagline": "Praxis offers labs the most diverse environments, SOTA hardware, and rigorous post processing to train their models", "body": "**Hey YC!**\n\nWe're Rohan, Dev, and Tommy, and we're building Praxis.\n\n**Praxis turns real-world work into training data for robots.**\n\n\n\n**What we do**\n\nRobots learn to act in the physical world by watching humans do the work. We partner with businesses to capture egocentric human demonstration data, the first-person video of how tasks actually get done in factories, warehouses, homes, and other physical work environments, then supply it to the labs and humanoid teams training the next generation of robots. Robotics data is our focus, but since our collection infrastructure overlaps, we're happy to explore other multimodal data too.\n\n**The problem**\n\nRobotics is missing its internet-scale training corpus. LLMs learned from the web, books, code, and the accumulated record of human thought. Robots need the equivalent record of human action: how people pick, pack, clean, assemble, inspect, cook, use tools, recover from mistakes, and move through messy physical environments.\n\nCollecting some of this data is easy. Collecting enough of the right kind is brutally hard, for two reasons.\n\nFirst, diversity is capped by access. Models only generalise if they've seen the full range of real environments and operators, from a one-person shop to a national-scale industrial operation. Getting in the door across that entire spectrum, and across continents, is a physical-world access problem that most teams can't solve.\n\nSecond, it takes two rare skills at once. You have to know what frontier labs need at a technical level, and be able to run demanding physical operations across thousands of real sites, standardising capture and turning fragmented workflows into model-ready data. Few teams have both halves.\n\n**Our solution**\n\nWe already run collections across 150 environment types on 5 continents, with diversity from mom-and-pop stores to conglomerates accounting for 3% of a nation's GDP, and a reachable network of 60,000 workers. Depending on the task, we capture egocentric video, stereo-RGB, IMU, wrist-camera views, glove kinematics, haptics, and dense annotations. Everything is delivered calibrated, synced, pose-tracked, QA'd, and PII-scrubbed. We build and deploy custom capture hardware quickly because part of the team operates out of Shenzhen.\n\n**Why us**\n\nPraxis is built to do both hard things at once. We pair robotics research depth (IEEE, NASA JPL) with the ability to actually run demanding operations in the real world, from Fortune 500 supply chains to Parliaments, plus a Shenzhen hardware team. That combination- technical understanding of what labs need plus real physical-world access at scale- is what lets us deliver diverse, model-ready data at a cost and breadth that is irreplaceable.\n\n**Our asks**\n\n1. Intros to anyone working on robotics, embodied AI, or world-models, especially data and research leads. If you need other forms of multimodal data, we're happy to explore that too.\n2. If you run a business with interesting physical work (logistics, manufacturing, food, retail, or lab work) and would like to monetize it, we'd love to talk.\n\nYou can reach us at [founders@praxisrobotics.io](mailto:founders@praxisrobotics.io) and [rohan@praxisrobotics.io](mailto:rohan@praxisrobotics.io)[ https://www.praxisrobotics.io](https://www.praxisrobotics.io)", "url": "https://www.ycombinator.com/launches/STf-praxis-ai-training-data-for-robotics", "votes": 11, "created_at": "2026-08-06T22:30:00.618Z" } ] }, { "id": 33144, "name": "Grip", "slug": "grip", "former_names": [], "small_logo_thumb_url": "https://bookface-images.s3.amazonaws.com/small_logos/a6be2671c5d4e3adfe916a4a59aad4680c7cc9d3.png", "website": "https://www.griprobotics.ai", "all_locations": "Zürich, ZH, Switzerland", "long_description": "Grip is a robotics company building intelligent robotic systems for waste and recycling. We are starting with organic waste, where removing contaminants from messy, constantly changing streams remains difficult and highly manual. Our technology combines learning-based manipulation and custom hardware to identify and pick objects, learning from real-world operating data, including every successful and failed grasp, to become more reliable and adaptable over time. This helps waste operators reduce manual sorting, recover more valuable material, lower contamination, and deploy automation flexibly across facilities. Over time, Grip aims to expand beyond organics and build the intelligent manipulation layer for trash in general, making waste processing and the circular economy more efficient and scalable.", "one_liner": "Robots for waste manipulation", "team_size": 4, "industry": "Industrials", "subindustry": "Industrials -> Manufacturing and Robotics", "launched_at": 1785254669, "tags": [ "Robotic Process Automation", "Sustainability" ], "tags_highlighted": [], "top_company": false, "isHiring": false, "nonprofit": false, "batch": "Summer 2026", "status": "Active", "industries": [ "Industrials", "Manufacturing and Robotics" ], "regions": [ "Switzerland", "Europe" ], "stage": "Early", "app_video_public": false, "demo_day_video_public": false, "app_answers": null, "question_answers": null, "url": "https://www.ycombinator.com/companies/grip", "api": "https://devasheeshg.github.io/yc-api/batches/summer-2026/grip.json", "year_founded": 2026, "city": "Zürich", "country": "CH", "linkedin_url": "https://www.linkedin.com/company/griprobotics", "x_url": "", "fb_url": null, "cb_url": null, "github_url": null, "logo_url": "https://bookface-images.s3.amazonaws.com/small_logos/a6be2671c5d4e3adfe916a4a59aad4680c7cc9d3.png", "app_video_url": null, "dday_video_url": null, "company_photos": [ "https://bookface-images.s3.us-west-2.amazonaws.com/attachments/f0ba95cdc010ecb5489de28455277fd61e5501df.png" ], "primary_partner": { "name": "Jared Friedman", "url": "https://www.ycombinator.com/people/jared-friedman" }, "founders": [ { "user_id": 3678212, "full_name": "Jonas Gruetter", "title": "Founder/CEO", "founder_bio": "Former robotics researcher, published at ICRA. Ex-Gravis Robotics, Robotic Systems Lab, and Ursa Mining. Now building robots to solve fundamental societal problems.", "is_active": true, "linkedin_url": "https://www.linkedin.com/in/jonasgrutter/", "x_url": "", "avatar_thumb_url": "https://bookface-images.s3.us-west-2.amazonaws.com/avatars/81ff3bd631bc4b7206041d530bc182993fddd8fc.jpg" }, { "user_id": 3678142, "full_name": "Stanislaw Piasecki", "title": "Founder/CTO", "founder_bio": "Co-founder at Grip. Previously founding engineer at URSA Mining, building autonomous excavators with custom control circuits, RL simulators, and multi-sensor perception systems. Built autonomous interceptor drones at CDDS, autonomous race cars at AMZ and quadruped robotics at ANYbotics. MS in Computational Science & Engineering from ETH Zürich.", "is_active": true, "linkedin_url": "https://www.linkedin.com/in/stanislaw-piasecki/", "x_url": "", "avatar_thumb_url": "https://bookface-images.s3.us-west-2.amazonaws.com/avatars/253c9593fed9e59498847d4c158543099816d8f1.jpg" } ], "jobs": [], "news": [], "launches": [ { "id": 111317, "title": "Grip - Robotics for Waste Management", "tagline": "Robots that sort any waste and recover recyclables, learning from every grasp.", "body": "Hi everyone - Stan and Jonas here, founders of **Grip**. We build intelligent robots that sort _any_ waste, not just the clean, predictable objects existing systems are limited to. We're starting with plastic contamination in organic waste.\\\n\\\n\n\n**The problem**\n\nWe didn't start with a thesis, we found it on sorting lines, after visiting waste facilities across several countries:\n\n* **Contamination is existential.** Plastic that organic-waste facilities can't remove gets shredded, processed, and spread onto agricultural soil. Operators spend up to 30% of operating costs fighting it and still fall short. Even Germany's 0.1% limit for plastic in finished compost is a tacit admission that full removal isn't possible with today's tools.\n* **The labor force is aging and shrinking.** The picking work is unpleasant and turnover is brutal. Operators increasingly aren't choosing between people and robots. They're choosing between robots and nobody.\n* **Dangerous items cause fires.** Batteries, propane tanks, and vapes slip through and ignite during handling and shredding: 448 reported fires at US and Canadian waste facilities in 2025, \\~$2.5B in damage, and insurers are walking away. At the same time trash sorting is sitting in the top 5 of US deadliest occupations.\n\n**What we built**\n\nThe hard part isn't the gripper. It's that waste never looks the way you planned. Objects arrive crushed, folded, tangled, half-buried, and moving. Put today's sorting robots in front of this and they break.\n\nWe're building one manipulation system that picks whatever shows up, at the front of the line where the material is most chaotic and existing automation stops. That means chaotic scene understanding, robust grasp planning, and hardware built for real waste streams: our custom three-way gripper is designed to maximize throughput, taking out as much contamination per cycle as possible.\n\nOur rugged hardware matters, but intelligence is the product. Every pick generates data no other system ever encounters: what the robot saw, what it tried, how the material responded, whether it worked.\n\n![uploaded image](/media/?type=post&id=111317&key=user_uploads/3678142/bc867fdc-9225-4b85-83f9-f83433465d5b)\n\n**Why now**\n\nWaste policy has moved from vague ambition to enforceable deadlines:\n\n* Separate bio-waste collection is now mandatory across the EU. More material, and more contamination, every year.\n* New EU fertiliser rules just cut permitted impurities in compost and digestate, with national limits like Germany's stricter still.\n* The US is real: California mandated a 75% cut in landfilled organics, missed it, and enforcement is ramping. At the same time, landfills near major cities are filling up, and that pressure is pushing more and more US municipalities to collect organic waste separately. More material, more contamination, more facilities that need to sort it.\n\nMore than 6,500 biowaste plants and thousands of recycling facilities across Europe and North America face this with no robot that can touch it. That's Grip.\n\n**Why us**\n\nWe've built autonomous race cars, quadrupeds, drones, and excavators, robots that operate in dirt, weather, and chaos rather than controlled labs. Grip is that experience applied to an industry where physical intelligence removes a real bottleneck.\n\n**Our ask**\n\nWe'd love intros to:\n\n* Owners and operators of composting, anaerobic-digestion, and organic-waste facilities\n* Owners and operators of MRFs and recycling facilities\n* Regulators and experts shaping contamination limits and recycling standards\n\nIf someone runs a facility and will show us their hardest sorting problem, we'll get on a plane.\n\n[**griprobotics.ai**](http://griprobotics.ai)", "url": "https://www.ycombinator.com/launches/SxR-grip-robotics-for-waste-management", "votes": 15, "created_at": "2026-08-19T13:00:00.941Z" } ] }, { "id": 36775, "name": "Asim", "slug": "asim-group", "former_names": [], "small_logo_thumb_url": "https://bookface-images.s3.amazonaws.com/small_logos/a3d66de7c8d8809683c96657196dfeabd35eb9bc.png", "website": "https://www.asim-group.com", "all_locations": "San Francisco, CA, USA", "long_description": "", "one_liner": "Enable robots to operate safely in the real world.", "team_size": 4, "industry": "Industrials", "subindustry": "Industrials -> Manufacturing and Robotics", "launched_at": 1790058078, "tags": [ "Deep Learning", "Robotic Process Automation", "Robotics" ], "tags_highlighted": [], "top_company": false, "isHiring": false, "nonprofit": false, "batch": "Fall 2026", "status": "Active", "industries": [ "Industrials", "Manufacturing and Robotics" ], "regions": [ "United States of America", "America / Canada" ], "stage": "Early", "app_video_public": false, "demo_day_video_public": false, "app_answers": null, "question_answers": null, "url": "https://www.ycombinator.com/companies/asim-group", "api": "https://devasheeshg.github.io/yc-api/batches/fall-2026/asim-group.json", "year_founded": null, "city": "San Francisco", "country": "US", "linkedin_url": "https://www.linkedin.com/company/asim-group", "x_url": "https://x.com/asim_group", "fb_url": null, "cb_url": null, "github_url": null, "logo_url": "https://bookface-images.s3.amazonaws.com/small_logos/a3d66de7c8d8809683c96657196dfeabd35eb9bc.png", "app_video_url": null, "dday_video_url": null, "company_photos": [], "primary_partner": { "name": "Nicolas Dessaigne", "url": "https://www.ycombinator.com/people/nicolas-dessaigne" }, "founders": [ { "user_id": 3703237, "full_name": "Douglas Tilley", "title": "Founder/CEO", "founder_bio": "Founder at Asim. PhD candidate in AI with Robotics at the University of Bath. Previously ML researcher at National Institute of Informatics (Tokyo) and Tianjin University. 3x Doctoral Recognition Award Winner. Researcher on Safe Human Robot interaction with multiple papers in Hierarchical Machine Learning, Robot Learning and Out-of-Distribution detection.", "is_active": true, "linkedin_url": "https://www.linkedin.com/in/douglas-tilley-uob2021phdart-ai/", "x_url": "https://x.com/DougTilley7", "avatar_thumb_url": "https://bookface-images.s3.us-west-2.amazonaws.com/avatars/a00c9a02b2cc8d085dd83366ea3e34a737bee5ad.jpg" }, { "user_id": 3700982, "full_name": "Raphaël Lafargue", "title": "Founder/CTO", "founder_bio": "Co-founder at Asim. ex-Senior Research Scientist @ Sony / PhD in AI @ IMT Atlantique/University of Adelaide / M.Sc. Physics @ ENS Paris-Saclay.", "is_active": true, "linkedin_url": "https://www.linkedin.com/in/raflaf", "x_url": "https://x.com/raflaf_", "avatar_thumb_url": "https://bookface-images.s3.us-west-2.amazonaws.com/avatars/f5c24f0c7c900d3821cd33af3b64d56c25d22476.jpg" }, { "user_id": 2289752, "full_name": "Tristan Lecourtois", "title": "Founder/COO", "founder_bio": "Founder at Asim (YC F26). Ex NASA JPL, MS from ENS Paris-Saclay.", "is_active": true, "linkedin_url": "https://www.linkedin.com/in/tristan-lecourtois", "x_url": "https://x.com/tristanlecourt", "avatar_thumb_url": "https://bookface-images.s3.us-west-2.amazonaws.com/avatars/b4137cc0d6140061e2b3763a49e6932ff3d38c7e.jpg" }, { "user_id": 3700934, "full_name": "Maëlic Neau", "title": "Founder", "founder_bio": "Founder at Asim (YC F26). PhD in Computer Science from Flinders University. Two times world champion of the RoboCup@Home international competition. Published at ICCV, oral presentation at BMVC 2025. Previously Visiting Scholar at National Institute of Informatics.", "is_active": true, "linkedin_url": "https://www.linkedin.com/in/maelicneau", "x_url": "https://x.com/MaelicN", "avatar_thumb_url": "https://bookface-images.s3.us-west-2.amazonaws.com/avatars/ae1fe59a371a12d1340c627534559abaea4f4868.jpg" } ], "jobs": [], "news": [], "launches": [] } ]