Rohit Patel, who spent eight years working on AI at Meta, has departed the company and launched a startup called Ndoc, TechCrunch has learned. Patel was previously a director at Meta Superintelligence Labs.
He presents Ndoc as an AI-native replacement for the fragmented mix of tools many teams use today — tools such as Excel, Word, PowerPoint, Tableau, Replit and ad-hoc ChatGPT copy-paste workflows.
Three-part structure: Stage, Narrative, Model
Ndoc combines three components in a single interface:
- Stage: where key presentations live — slide decks, dashboards, simulations or mini apps.
- Narrative: the analyses, numbered citations, generated figures and the explanatory “why” behind the numbers.
- Model: the underlying datasets and Python scripts that produce the outputs.
That structure reflects Patel’s background: at Meta he focused on reinforcement learning and on evaluating whether AI outputs are trustworthy. Patel argues that if AI is producing spreadsheets and slide decks used for decisions, users should be able to inspect the work — for example by locating the Python scripts that generated a document.
"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."
Workflow, memory and imports
In practice, users can upload a dataset, ask Ndoc to run an analysis and produce a narrative about the findings. Documents can carry specific context such as audience, tone and memory. Patel says the system maintains memory across your personal use, that particular document, your team and your company.
Recognizing the difficulty of changing entrenched workflows, Patel built connectors so customers can import existing data from the tools they already use.
Model choices and operating costs
Patel says Ndoc is built and fine-tuned using a mix of open-source models, which he believes makes it considerably cheaper to run than alternatives built on cutting-edge proprietary models. Early trial users, he said, were less excited about cost savings and more interested in seeing their entire end-to-end workflow in one place.
A larger technical aim is to reinvent the file format. Documents can be exported as a single ".ndoc.html" file that opens in any browser; Patel intends to open-source the format so other developers can build apps that work with it rather than locking users into a single vendor.
Why he left Meta and why speed matters
Patel cites the desire to open-source the format and to move quickly as reasons for leaving Meta. He argued that at a large company, arriving at a consensus on a new open format would require the input of many people and thus be slow. "Humans can only move at human pace, whereas… I can move at the pace of what AI can do," he said.
Competitive landscape and current stage
Trying to become the enterprise platform where work happens is a big bet for a three-person, unfunded startup. Several large AI labs — OpenAI, Anthropic, Meta, SpaceXAI and Google — have released enterprise AI tools that integrate with Google Workspace, Microsoft 365 and other workplace apps. OpenAI also recently launched Space, a shared workspace inside ChatGPT that more directly competes with traditional productivity suites.
Patel acknowledges those products could be threats but argues that many labs attach AI to existing spreadsheets in ways he finds superficial. "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."
He says he understands the challenge of changing habits, which is why Ndoc supports importing data from current tools.
Launch plan and monetization
As of Tuesday, Ndoc is running an enterprise pilot and opening an alpha to a limited number of customers who join a wait list. Signing up is free; customers will pay for usage and tokens. Patel hopes to ship a beta in the coming weeks and aims for a full launch next year.
Patel has set a high bar for adoption before raising outside capital: he did not want to take funding until he had proof people loved the product. "I didn’t want to take money until I had some proof that people love the product," he said.
For Patel, moving fast 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."
Ndoc’s immediate future depends on whether the product can win enough user enthusiasm and enterprise traction to fund further growth and fend off competition.
