After eight years building AI at Meta, Rohit Patel is leaving to launch Ndoc, a startup he says could replace Microsoft Office and Google Workspace with an AI-native workspace. The former director at Meta Superintelligence Labs is pitching the product as more than a layer of AI on top of existing apps: he wants it to become the place where enterprise work actually happens.
Rohit Patel’s exit comes amid broader Meta departures
TechCrunch reports that Patel is the latest high-profile departure from Meta, following exits by Chief Global Affairs Officer Joel Kaplan, AI researcher Andrew Tulloch, and Head of Product for AI-for-Work Emily Dalton Smith. His move ends an eight-year stretch at the company, where he focused on AI development and, more specifically, on reinforcement learning and evaluating whether AI output can be trusted.
That background shows up directly in Ndoc’s design. Patel has built the startup around the idea that if AI is going to generate spreadsheets, presentations, and decision-making documents, users need a way to inspect the underlying work instead of accepting results blindly.
Ndoc aims to unify work across spreadsheets, slides, code and chat
Patel says the product is meant to replace the fragmented workflow many teams now spread across Excel, Word, PowerPoint, Tableau, Replit and what he called “ChatGPT copy-paste.” Rather than forcing users to bounce between separate tools, Ndoc is designed as a shared workspace that colleagues can access even if they are not signed up for the service.
The platform is organized into three main parts:
- Stage — the home for presentations of key findings, including slide decks, dashboards, simulations and mini apps.
- Narrative — the area for analysis, numbered citations, generated figures and the explanation behind the numbers.
- Model — the section that stores the underlying datasets and Python scripts.
In practice, users can upload a dataset and ask Ndoc to analyze it and write up the findings. They can also add context about audience, tone and memory so the output is tailored to the work at hand.
Transparency is part of the product pitch
Patel says the system is built so people can verify how results were produced. Because the documents include the Python scripts used to generate them, users can inspect formulas and logic rather than treating the output as a black box.
“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.”
That emphasis on trust reflects his Meta experience. He argues that if AI is going to generate the documents companies use to make decisions, the system should make auditing those documents straightforward.
Built on open source models and designed to cost less
Patel says Ndoc is built and fine-tuned using a mix of open source models. Because the platform was created from scratch, he says it is considerably cheaper to operate than products built on frontier AI models.
But cost savings are not what first drew attention from trial users, according to Patel. He said early testers cared more about having their entire workflow in one place than about the underlying economics. That response appears to be shaping how he positions the startup as it moves toward a wider launch.
He wants to reinvent the file itself
Patel’s ambition goes beyond apps and workflows. He wants to rethink the document format itself by allowing exports as a single “.ndoc.html” file that can open in any browser.
He also plans to open-source the format so other developers can build compatible apps without being locked into Ndoc. Patel contrasted that goal with the long-standing dependence many businesses have on Microsoft and Google formats, saying the open approach is meant to make documents more portable and less tied to one vendor.
He said leaving Meta was partly necessary because moving such a format forward inside a large company would require agreement across “hundreds of people and opinions” on how it should work. Patel wanted to move faster than that.
A crowded race to own AI-powered office work
Ndoc enters a competitive field. OpenAI, Anthropic, Meta, SpaceXAI and Google have all introduced enterprise AI tools that connect to Google Workspace, Microsoft 365 and other workplace apps so models can work directly inside them. OpenAI also launched Space last month, a shared workspace inside ChatGPT that looks more like a direct rival to office software.
Patel acknowledges that approach, but says the industry is aiming too low by merely attaching AI to existing spreadsheets and documents. He argues that the opportunity is to rethink the workflow rather than automate a few cells at a time.
“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.”
Next steps: alpha access now, beta soon
Ndoc is already in an enterprise pilot and is opening an alpha version to a limited number of customers on its wait list. It is free to sign up, but users will pay for usage and tokens. Patel expects a beta release in the coming weeks and says a full launch is planned for next year.
He has not raised funding yet and says he wants proof that people love the product before taking money. “I didn’t want to take money until I had some proof that people love the product,” he said.
Patel’s approach is bluntly pragmatic: ship quickly, learn from usage and keep moving. “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.”
Source: Original report
Was this helpful?
Explore more: AI Automation Services More Business & Startups Tech News
Last Modified: October 27, 2026 at 10:31 pm
2 views
