After eight years developing AI for Meta, director Rohit Patel is departing the company, TechCrunch has exclusively learned. The former director at Meta Superintelligence Labs has launched Ndoc, a startup he is pitching as an AI-native replacement for Microsoft Office and Google Workspace.
Rohit Patel (Credits: TechCrunch)
Meta AI Director Rohit Patel Leaves for Ndoc Startup
Patel is the latest in a string of high-profile departures from Meta, following the exits of Chief Global Affairs Officer Joel Kaplan, AI researcher Andrew Tulloch, and Head of Product for AI-for-Work Emily Dalton Smith. His new venture aims far beyond bolting AI onto existing tools: Ndoc is designed to replace the fragmented way people work across Excel, Word, PowerPoint, Tableau, Replit, and “ChatGPT copy-paste” with one unified workspace colleagues can access even without signing up.
Ndoc brings three components together in one layout. The Stage tab holds presentations of key findings — slide decks, dashboards, simulations, or mini apps. The Narrative tab carries analyses, numbered citations, generated figures, and the “why” behind the numbers. The Model tab houses the underlying datasets and Python scripts.
Ndoc: An AI-Native Office Replacement With Verifiable Work
The design reflects Patel’s Meta background in reinforcement learning and in evaluating whether AI output is trustworthy. His argument: if AI is building the spreadsheets and slide decks companies use to make decisions, someone should be able to check its work. Users can inspect the Python scripts behind documents, or ask the AI chat to explain what a script does and which line holds a formula. “There’s nothing hidden behind the curtain here,” Patel said.
In practice, users upload a dataset and ask Ndoc to analyze it and write a narrative about the findings, adding document-specific context such as audience, tone, and memory. The platform maintains memory across the person, the document, the team, and the company, and is built and fine-tuned on a mixture of open-source models — which Patel says makes it considerably cheaper to run than alternatives built on frontier models.
Patel’s larger ambition is to reinvent the file itself: documents export as a single “.ndoc.html” file that opens in any browser, a format he plans to open-source so other developers can build compatible apps without being locked into Ndoc the way users have been locked into Microsoft or Google for years. That open-source ambition was partly why he left Meta, where agreeing on a format would require “hundreds of people and opinions.”
It is a huge bet for a founder with a team of three and no funding, especially as OpenAI, Anthropic, Meta, SpaceXAI, and Google all push enterprise AI tools — and OpenAI launched Space, a shared workspace inside ChatGPT, just last month. But Patel is betting that end-to-end work in one place, with verifiable AI, is what users actually want.
Also Read: ContextGem 1.0: Extract Data From Documents With LLMs and Decision Models | Data Center Backlash Is Uniting America Ahead of the Midterms
Source: TechCrunch — read the full story there.
