Former Meta AI Director Rohit Patel launches Ndoc to challenge Office and Workspace

Rohit Patel, who spent eight years developing AI at Meta, has launched Ndoc, an AI-native workspace he says is designed to replace Microsoft Office and Google Workspace. The startup, reported by TechCrunch, is aiming to bundle writing, analysis, presentations and code-backed workflows into a single product instead of forcing users to juggle separate tools and copy-paste between them.

What is Ndoc?

Ndoc is Patel’s bet that office software can be rebuilt around AI from the start rather than layered with assistants on top. According to TechCrunch, the product is built around three core tabs — Stage, Narrative and Model — each meant to handle a different part of knowledge work.

How does the product work?

Patel’s pitch is that users should be able to move from a presentation to analysis to underlying code without leaving the workspace. The Stage tab is where presentations, dashboards, simulations or mini apps live. Narrative is where analyses, citations, figures and explanations appear. Model is where datasets and Python scripts are stored.

That structure is meant to reduce the fragmented routine many office workers know well: one file in Excel, a slide deck in PowerPoint, a draft in Word, data in Tableau, code in Replit and AI prompts copied into ChatGPT. Patel told TechCrunch that Ndoc is intended to collapse that chain into one environment.

Why did a former Meta AI leader leave to do this?

Patel left Meta after eight years, including a stint as director at Meta Superintelligence Labs, to build Ndoc from scratch. TechCrunch reports that he wants the product to move faster than the pace he had inside a large company, especially on the question of open source and file portability.

He framed that decision in blunt terms, saying,

“Humans can only move at human pace, whereas…I can move at the pace of what AI can do.”

That view appears to sit at the center of the startup’s strategy: use AI not as a feature but as the operating logic of the workspace itself.

Why it matters

The significance of Ndoc goes beyond another productivity app. If Patel’s approach works, it could challenge the basic structure of office software, where different tasks are split across distinct products and formats controlled by major vendors. For users, that could mean fewer handoffs, less manual reformatting and more direct access to the logic behind the output.

For the market, the more interesting threat is to lock-in. Open-source or portable file formats can make it easier to move work between systems, reducing dependence on a single suite. Patel says he plans a single exportable .ndoc.html file format and wants to open-source it, which would make portability a central selling point rather than a side feature.

For incumbents such as Microsoft and Google, the challenge is not just another competing interface. Ndoc is trying to reframe the category around end-to-end workflows, with transparency built into the product design. That is a different proposition from the typical AI add-on layered onto familiar apps.

Background: how we got here

Office software has long been organized around separate document, spreadsheet and presentation tools, even as workplace work itself has become more fluid. Over the past two years, AI labs and software companies have increasingly pushed into that space with connectors, copilots and shared workspaces meant to sit inside existing suites rather than replace them.

That shift has created a familiar pattern: AI helps draft, summarize or search, but the user still has to manage the final output across multiple applications. Patel’s argument is that this is the wrong place to stop. Instead of adding intelligence to fragmented work, Ndoc is trying to rebuild the work surface so the intelligence is native to the system.

TechCrunch says Patel’s background at Meta matters here because he has spent years inside large-scale AI development. He is not approaching the problem as a standalone app builder, but as someone who has worked on advanced AI infrastructure and now wants to apply those lessons to an office product.

What does Ndoc actually offer?

The clearest promise is inspection and transparency. Patel says users can inspect Python scripts to verify formulas and outputs. That is important because it suggests Ndoc is not just asking users to trust a black box; it is trying to make the chain from data to result visible.

Patel also says the product keeps memory across user, document, team and company context. In practical terms, that implies the workspace can learn from past work and carry context forward across projects rather than treating every file as a blank slate.

According to TechCrunch, Ndoc uses a mix of open-source models and was built from scratch. Patel also says that choice helps keep the startup’s costs below frontier-model-based alternatives. Lower cost matters because it may let a tiny team compete on efficiency even without the capital reserves of larger rivals.

By the numbers

How is Ndoc different from existing office tools?

Patel’s answer is that it tries to eliminate the handoff between creation, analysis and presentation. Rather than making a document that merely contains AI help, the product is organized so the AI-assisted workflow is the document.

The Stage tab can hold presentations, dashboards, simulations or mini apps. Narrative can explain the output, show the citations and display the figures. Model contains the data and code that support the analysis. That setup suggests a product aimed as much at analysts, technical teams and operators as at traditional office users.

Patel has also drawn a hard line on file format. He plans a single exportable .ndoc.html file and wants to open-source it. That would make the file itself part of the company’s strategy, not just the app that opens it.

He told TechCrunch that he left Meta partly to move faster on the open-source decision. That is notable because the file format could determine whether Ndoc becomes a silo or a bridge to other systems.

What is the commercial strategy?

For now, Ndoc is operating lean. TechCrunch says the company has a three-person team and no funding. It is already in an enterprise pilot and a limited alpha waitlist cohort, which suggests Patel is testing whether the workflow resonates before scaling the business.

Sign-up is free at this stage, with usage and token charges expected later, according to Patel. That pricing setup is consistent with a product that wants users in the door early and monetization later, once the workflow is established.

Patel expects a beta in the coming weeks and a full launch next year. Those milestones matter because they will show whether Ndoc can convert an ambitious product concept into something companies actually use inside daily work.

What happens next?

The next thing to watch is whether Ndoc can prove that its all-in-one design creates enough value to pull users away from established suites. The bigger the promise, the harder the proof requirement: users will need to see that the product is not just more elegant, but materially faster and more reliable.

There are also unresolved questions about distribution and funding. TechCrunch notes that Ndoc is still pre-scale, with no disclosed funding and a limited pilot footprint. If the beta draws interest, the company will still have to decide whether to raise capital before or after the launch.

Another open issue is adoption of the file format. If the planned .ndoc.html standard is genuinely open and easy to use, it could become the most strategically important part of the product. If not, Ndoc risks becoming just another isolated workspace.

And then there is competition. Larger labs and incumbents are already trying to embed AI inside the tools people use every day. Ndoc’s case is that those products are not bold enough. The coming months will show whether that critique resonates beyond a small pilot group.

Meta’s recent talent churn adds context

TechCrunch says Patel is one of several notable people to leave Meta recently, alongside Joel Kaplan, Andrew Tulloch and Emily Dalton Smith. That does not by itself explain Patel’s move, but it does place Ndoc inside a broader pattern of departures from one of the most closely watched AI companies.

For observers of the AI market, such exits are often as important as launches. They can signal where former insiders see room for a new product, a new process or a faster route to shipping. Patel’s wager is that the biggest opportunity is not another chatbot, but a workspace rebuilt around AI from the ground up.

Timeline

Bottom line

Ndoc is an early-stage, no-funding startup with a small team, but its ambition is unusually direct: replace the familiar office stack with an AI-native workspace that is transparent, portable and built around one connected workflow. Whether that turns into a real rival to Microsoft Office and Google Workspace will depend on the beta, the file format and whether users decide the promise of fewer tools is worth the switch.