After eight years developing artificial intelligence at Meta, Rohit Patel has departed the company to launch Ndoc, an AI-native productivity startup. The venture aims to challenge established enterprise software giants Microsoft Office and Google Workspace with a unified browser-exportable format.
Rohit Patel Departs Meta Superintelligence Labs After Eight Years
After eight years helping shape Meta’s artificial intelligence initiatives, Rohit Patel is stepping down from his role as director at Meta Superintelligence Labs. His departure marks the latest in a series of high-profile exits from the social media giant. Other departures include Chief Global Affairs Officer Joel Kaplan, AI researcher Andrew Tulloch, and Emily Dalton Smith, who served as the head of product for AI-for-work.
During his eight-year tenure, Patel focused heavily on reinforcement learning and model evaluation systems, leading AI evaluations and reinforcement learning work at the social media giant. Those frameworks became central to how Meta trains its models and evaluates whether AI outputs are trustworthy. Rather than staying inside Meta, however, Patel chose to exit and establish a startup without external funding after exploring ideas.
Humans can only move at human pace, whereas…I can move at the pace of what AI can do.
Rohit Patel, Founder of Ndoc
Ndoc Challenges Microsoft Office and Google Workspace With Browser-Exportable Files
Patel’s new venture, Ndoc, is designed to replace the fragmented toolchains that office workers typically juggle. Today’s workplace relies on an assortment of distinct applications, spreadsheets, presentations, notepads and conversation logs, all built prior to an era when artificial intelligence generates initial content. Ndoc combines presentations, analysis and underlying data in one workspace, uniting presentations, narratives and underlying models into one verifiable file.
The platform maintains memory across a user, a document, a team and a company, so context persists in a way that simple chat interfaces do not. The interface blends human edits, comments, and AI-generated changes, and is assembled from scratch using a mix of open source models to reduce running costs compared with systems built on frontier models. Connectors are available to import data from existing tools so migration can be gradual rather than forced. Documents can be exported as a single .ndoc.html file that opens in any browser, a format Patel plans to open-source.
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. There’s nothing hidden behind the curtain here.
Rohit Patel, Founder of Ndoc
The Enterprise AI Productivity Market and Established Incumbents
Challenging the productivity duopoly of Microsoft Office and Google Workspace involves confronting deeply embedded enterprise habits. Paid subscriptions for Microsoft Office exceed 345 million users, whereas Google Workspace caters to upwards of 9 million corporate clients. While major AI labs have rushed to connect their models directly into those established suites, Patel argues that simply bolting generative features onto legacy spreadsheets is an inadequate approach. Companies are basically attaching AI to spreadsheets and it fills your cells for you.
Enterprise Pilots
Ndoc is piloting the product with businesses. It is unclear if an unfinanced three-person startup can unseat entrenched office software, but the launch highlights a trend of Big Tech AI leaders starting their own companies to rethink enterprise workflows.