Rohit Patel, a director at Meta Superintelligence Labs who spent eight years developing AI for the company, has left Meta to launch his own startup. The company, called Ndoc, is pitched as an AI-native replacement for Microsoft Office and Google Workspace. TechCrunch first reported the departure.
Patel is the latest in a string of high-profile exits from Meta. Chief Global Affairs Officer Joel Kaplan, AI researcher Andrew Tulloch and Head of Product for AI-for-Work Emily Dalton Smith have also left the company recently.
Patel built Ndoc to be more than another AI tool bolted onto an existing workflow. His goal is to replace the scattered way people work today, which often means jumping between Excel, Word, PowerPoint, Tableau, Replit and what he calls “ChatGPT copy-paste.” In his vision, all of that happens in one workspace. Colleagues can access it even if they haven’t signed up for Ndoc themselves.
The product is organized into three areas. The Stage tab holds presentations of key findings, including slide decks, dashboards, simulations and mini apps. The Narrative tab contains analyses, numbered citations, generated figures and, in Patel’s words, the “why” behind the numbers. The Model tab stores the underlying datasets and Python scripts.
The design reflects Patel’s work at Meta, where he focused on reinforcement learning, a type of machine learning in which AI agents learn to make decisions through trial and error in pursuit of rewards. He also spent time evaluating whether AI output can be trusted. Patel argues that if AI is going to build the spreadsheets and slide decks companies use to make decisions, someone should be able to check its work. Users can find the Python scripts that generated their documents and review them.
“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.”
In practice, a user can upload a dataset and ask Ndoc to run an analysis and write a narrative about the findings. Users can also add context for each document, including the intended audience, tone and memory. Patel said the system keeps memory at several levels. “It maintains a memory across your person, that particular document, your team, and your company,” he said.
Ndoc is built and fine-tuned using a mix of open-source models. Because Patel developed the platform from scratch, he says it costs considerably less to run than alternatives built on frontier AI models. Cost, he said, was not what early trial users cared about most. “I thought that’s what people would be excited about, but people don’t care about that,” he said. “What people are excited about is that they love seeing all their end-to-end in one place.”
Patel’s broader ambition is to change the file itself. He says documents can be exported as a single “.ndoc.html” file that opens in any browser. He plans to open-source the format so other developers can build apps that work with it. The aim is to avoid the lock-in that many users have experienced with Microsoft and Google over the years.
Wanting to open-source the format was part of the reason Patel decided to leave Meta and work independently. At a company the size of Meta, he suggested, getting everyone to agree on an open-source format would require input from “hundreds of people and opinions on how the format should look.” Patel wanted to move faster. “Humans can only move at human pace, whereas…I can move at the pace of what AI can do.”
The bet is a large one. Ndoc has a team of three and no funding, and it is trying to become the platform where enterprise work happens. That puts it up against well-funded rivals. OpenAI, Anthropic, Meta, SpaceXAI and Google have all released enterprise AI tools that connect to Google Workspace, Microsoft 365 and other workplace apps, allowing their models to work directly inside those programs. Last month, OpenAI launched Space, a shared workspace inside ChatGPT that more directly competes with established productivity tools.
Patel acknowledges that Space could be a threat, but he thinks the major labs have the wrong approach. “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.”
The labs have a reason for bringing their models into tools people already use. Changing habits and replacing entire workflows is hard. Patel says he understands that, which is why Ndoc includes connectors that let customers import existing data from the tools they already use.
His standard for the product is personal. “My bar was that I’m not going to release anything I don’t love using,” he said. “I have very optimized AI workflows for everything I do. If I start using this tool, and I keep going back to my workflows, I haven’t built a good enough tool…I have a graveyard of tools that did not meet that bar.”
Patel also plans to hold off on fundraising until the product has traction. “I didn’t want to take money until I had some proof that people love the product,” he said.
As of Tuesday, Ndoc is running an enterprise pilot and opening an alpha version to a limited number of customers who join a waitlist. Signing up is free, but customers will pay for usage and tokens. Patel hopes to release a beta in the coming weeks, with a full launch planned for next year.
For Patel, speed matters more than launching a perfect product. “Execution is extremely cheap…so go execute on a suboptimal plan and worry about it later,” he said. “By the time you make the best plan, the execution will be done by somebody else.”
Cet article a été rédigé avec l'assistance de l'IA.
News Factory APP - actualités agentiques pour booster votre SEO et AEO.