Rohit Patel has left Meta after eight years to launch Ndoc, a startup combining AI-generated analysis, presentations and underlying data in one workspace.
Rohit Patel, formerly a director at Meta Superintelligence Labs, has departed Meta after eight years and launched Ndoc, a startup developing an AI-centered alternative to conventional office software. TechCrunch reported the departure on October 27, 2026. Patel says the product is intended to bring work now spread across spreadsheets, presentations, analytics tools and AI chats into a shared workspace.
Ndoc organizes projects into three areas. Stage presents outputs such as slide decks, dashboards, simulations and small applications. Narrative contains written analysis, citations and generated figures. Model exposes the supporting datasets and Python scripts. Users can upload data and ask the system to analyze it and explain the results; they can also supply context such as a document’s audience and tone. Patel says the product maintains memory at the individual, document, team and company levels.
A central design choice is to make the work behind generated materials inspectable. Users can access the scripts used to create documents and ask the AI to explain calculations, allowing them to check formulas rather than rely solely on a finished spreadsheet or presentation. Patel’s prior Meta work included reinforcement learning and evaluating the reliability of AI outputs. He says Ndoc is built and fine-tuned with a mix of open-source models, which he says costs less to operate than systems based on frontier models.
The company is also developing an export format, .ndoc.html, that Patel plans to open-source so other developers can build compatible applications. Ndoc is conducting an enterprise pilot and has opened a limited alpha to people on its waitlist. Joining is free, but customers will pay for usage and tokens. Patel said he hopes to release a beta in the coming weeks and plans a full launch next year. The startup has a team of three and had not raised funding at the time of the report.
