Patel, who focused on reinforcement learning and AI trustworthiness at Meta, announced the startup exclusively to TechCrunch. Ndoc combines three tabs: Stage for presentations and dashboards, Narrative for analyses and citations, and Model for underlying datasets and Python scripts. The design prioritises verifiability, users can inspect the code behind generated documents, and an AI chat can explain it in plain language.
“Even if you don't know any Python, you can ask [the AI chat] to explain, ‘Hey, what does this script do? Which line has my formula?’ And you can verify that formula,” Patel said. “There’s nothing hidden behind the curtain here.”
Ndoc is built on a mix of open-source models and fine-tuned from scratch, which Patel says keeps operating costs well below those of rivals relying on frontier AI models. The platform maintains a persistent memory across the user, document, team and company.
Patel's larger ambition is to reinvent the file format itself. Documents export as single .ndoc.html files viewable in any browser, a format he plans to open source so other developers can build compatible tools, avoiding the lock-in he says Microsoft and Google impose. He chose to leave Meta partly because agreeing on such a format inside a large company would involve “hundreds of people and opinions.”
Ndoc faces stiff competition. OpenAI, Anthropic, Meta, SpaceXAI and Google have all launched enterprise AI tools that connect to existing office suites. Last month OpenAI released Space, a shared workspace inside ChatGPT that directly targets the same market. Patel acknowledged Space could be a threat, but argued that most labs are approaching the problem wrong. “Companies are basically attaching AI to spreadsheets and it fills your cells for you,” he said. “It just sort of kills my soul. This is like if you just figured out electricity, and then you decided the best use case is to put lights on my bulls that till my soil. No. Get a tractor.”
Patel has a team of three and no funding yet, setting a bar of customer love before raising money. As of Tuesday, Ndoc is running an enterprise pilot and opening an alpha version through a wait list. It is free to sign up; users will later pay for usage and tokens. A beta is expected “in the coming weeks” and a full launch next year. “Execution is extremely cheap… so go execute on a suboptimal plan and worry about it later,” Patel said. “By the time you make the best plan, the execution will be done by somebody else.”
Patel is the latest in a series of high-profile Meta departures, including Chief Global Affairs Officer Joel Kaplan, AI researcher Andrew Tulloch, and Head of Product for AI-for-Work Emily Dalton Smith.