A senior figure from Meta's artificial intelligence division has stepped away from the company after nearly a decade to take on one of the most entrenched duopolies in enterprise software. Rohit Patel, who served as a director at Meta Superintelligence Labs, has left to found a startup called Ndoc - a platform he describes as a ground-up, AI-native alternative to productivity suites like Microsoft 365 and Google Workspace.

Patel's departure adds to a growing list of high-profile exits from Meta in recent months. Among those who have left the company are Chief Global Affairs Officer Joel Kaplan, AI researcher Andrew Tulloch, and Emily Dalton Smith, who led product development for Meta's AI-for-Work initiative. The string of departures reflects a broader pattern of senior talent leaving large technology firms to pursue independent ventures in the AI space.

One Workspace to Replace Many

The central idea behind Ndoc is not simply to add another AI feature on top of an existing suite of tools. Rather, Patel's vision is to eliminate the fragmented experience that most knowledge workers face when juggling applications like Excel, Word, PowerPoint, Tableau, Replit, and various AI chatbots. In its place, Ndoc offers a single, unified workspace that team members can access together - even those who have not registered for the platform themselves.

The product is structured around three distinct tabs, each serving a specific function. The Stage tab is where key findings are presented - think slide decks, dashboards, simulations, or lightweight interactive applications. The Narrative tab holds the analytical layer: written explanations, numbered citations, auto-generated figures, and the reasoning that sits behind the data. The Model tab stores the raw materials - datasets and the Python scripts used to process them.

This three-part architecture is a direct reflection of Patel's professional background. During his eight years at Meta, he worked extensively on reinforcement learning - a branch of machine learning in which AI systems improve their decision-making through repeated trial, error, and reward signals - and on assessing whether AI-generated outputs can be trusted. That experience shaped his belief that if AI is going to produce the spreadsheets and presentations that businesses rely on to make decisions, those outputs need to be auditable.

Transparency as a Core Design Principle

One of Ndoc's distinguishing features is its approach to transparency. Users can inspect the Python scripts that generated any document within the platform. Patel argues that this matters even for people without a technical background.

"Even if you don't know any Python, you can ask the AI chat to explain what a script does, which line has your formula - and you can verify that formula. There's nothing hidden behind the curtain here," Patel said.

In practical terms, a user can upload a dataset and instruct Ndoc to run an analysis and produce a written narrative summarizing the findings. The platform also allows users to set document-specific parameters, including the intended audience, the desired tone, and persistent memory. According to Patel, the system retains context at multiple levels - for the individual user, the specific document, the team, and the broader organization.

On the technical side, Ndoc is built on a combination of open-source AI models rather than relying on frontier proprietary systems. Patel says this architectural choice makes the platform considerably cheaper to operate than competing products. However, he noted that cost efficiency has not been the selling point that resonated most with early users.

"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 work end-to-end in one place."

Reinventing the File Format

Beyond the interface and workflow, Patel has a more fundamental ambition: to rethink the document itself. Ndoc allows documents to be exported as a single ".ndoc.html" file that can be opened in any standard web browser - no proprietary software required. Patel intends to open-source this format, enabling other developers to build applications that are compatible with it. His goal is to offer an alternative to the long-standing lock-in that has kept so many organizations tethered to Microsoft or Google's ecosystems.

The desire to pursue open-source development was, in part, what prompted him to leave Meta and work independently. At a company of that scale, he explained, agreeing on an open-source file format would inevitably involve input from a large number of stakeholders, each with their own perspective on how the specification should be designed.

Patel wanted to move without that overhead. "Humans can only move at human pace, whereas I can move at the pace of what AI can do," he said.

A Crowded and Well-Funded Arena

The ambition to become the default platform for enterprise work is a substantial wager for a founder operating with a team of three and no external funding - particularly at a time when some of the most well-resourced AI organizations in the world are competing for the same territory.

Companies including OpenAI, Anthropic, Meta, and Google have all introduced enterprise AI tools that connect directly to existing productivity platforms like Google Workspace and Microsoft 365, allowing their models to operate within those familiar environments. OpenAI recently launched a feature called Spaces - a shared workspace built into ChatGPT that positions itself as a more direct competitor to established collaboration tools.

Patel acknowledges that Spaces could pose a competitive challenge, but he takes issue with the broader approach that AI labs have adopted when entering the productivity software market.

"Companies are basically attaching AI to spreadsheets and it fills your cells for you. 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."

There is, of course, a practical rationale for the approach that larger AI companies have taken. Changing established habits and displacing entire workflows is notoriously difficult. Patel says he understands this dynamic, which is why he built data connectors into Ndoc that allow customers to import existing information from the tools they currently use. The aim is to lower the barrier to adoption rather than demanding a complete break from existing systems on day one.

A High Bar Before Seeking Investment

Patel has set a personal standard for what Ndoc needs to achieve before he considers it ready - and that standard applies to his own usage as much as anyone else's.

"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 old workflows, I haven't built a good enough tool. I have a graveyard of tools that did not meet that bar."

He applies a comparable standard to fundraising. Rather than seeking investment based on a pitch, Patel wants to demonstrate genuine user enthusiasm first. "I didn't want to take money until I had some proof that people love the product," he said.

As of this week, Ndoc has entered an enterprise pilot phase and is opening an alpha version to a limited number of users who join a waitlist. Access is free to sign up for, though customers will eventually pay based on usage and token consumption. Patel expects to release a beta version within the next few weeks, with a full public launch targeted for next year.

Speed Over Perfection

For Patel, the priority is momentum over polish. He is operating with the conviction that in the current AI landscape, the cost of execution has dropped dramatically - and that waiting for the optimal plan is a risk in itself.

"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."

Whether a small, unfunded team can carve out meaningful ground against incumbents with billions in resources remains to be seen. But Patel's background in AI research, his focus on transparency and auditability, and his intention to open-source a new document format give Ndoc a distinct set of ideas - even if the road ahead involves competing against some of the most powerful companies in the technology industry.