Exclusive: Meta director Rohit Patel departs after 8 years to take on Microsoft Office Suite
Rohit Patel, who spent eight years on AI at Meta, has left to start Ndoc, a three-person startup building an AI-native workspace meant to replace Word, Excel, PowerPoint, and similar tools. It's in alpha with a waitlist, and Patel is betting that a readable, open file format and verifiable AI output will beat the big labs' attempts to bolt AI onto existing apps.
- Ndoc combines slides, analysis, and datasets in one layout (Stage, Narrative, Model tabs), and keeps the Python scripts behind every number so users can check the work.
- Patel plans to open-source a single ".ndoc.html" file format that opens in any browser, aiming to break the lock-in of Microsoft and Google files.
- The startup has three people and no funding, is free to sign up but charges for usage, and plans a beta in coming weeks with a full launch next year. OpenAI's new Space product is a direct competitor.
Rohit Patel, who spent eight years building AI at Meta and most recently directed work at Meta Superintelligence Labs, has left to start Ndoc, which he pitches as an AI-native replacement for Microsoft Office and Google Workspace. The company has a team of three and no outside funding. Patel's exit is one of several recent departures from Meta, following Chief Global Affairs Officer Joel Kaplan, AI researcher Andrew Tulloch, and Head of Product for AI-for-Work Emily Dalton Smith. The pitch is not to bolt AI onto Excel, Word, and PowerPoint but to collapse the mess of tools people juggle, including Tableau, Replit, and copy-pasting into ChatGPT, into one workspace. Colleagues can view outputs even if they don't have an account.
Ndoc organizes work into three tabs. Stage holds presentations of findings such as slide decks, dashboards, and mini apps. Narrative holds the analysis itself, with numbered citations and generated figures. Model holds the underlying datasets and Python scripts. Patel's background in reinforcement learning and in checking whether AI output can be trusted shapes the design. His argument is that if AI is building the spreadsheets that drive business decisions, people should be able to audit them. Users can read the scripts, or ask the built-in AI to explain which line holds a given formula. He also says the platform is built on open-source models, which he claims makes it much cheaper to run than products built on frontier models. Early testers, though, cared more about having everything in one place than about the cost.
Patel's bigger goal is to change the file format itself. Documents export as a single ".ndoc.html" file that opens in any browser, and he plans to open-source the format so others can build on it without being locked in the way Microsoft and Google users have been. He left Meta partly because getting a format through a large company would mean endless debate, and he wants to move at what he calls the pace of AI. The competition is steep. OpenAI, Anthropic, Meta, SpaceXAI, and Google already offer enterprise tools that plug into Workspace and Microsoft 365, and OpenAI launched Space last month as a shared workspace inside ChatGPT. Patel says the labs are attaching AI to spreadsheets in the wrong way, and he is building connectors so teams can import their existing data. Ndoc is running an enterprise pilot and an alpha for waitlisted customers, which is free to sign up for but charges for usage and tokens. A beta is expected within weeks and a full launch next year. He says he won't raise money until users prove they love the product.