With three developers and open source, former Meta director launches Ndoc to challenge Microsoft and Google suites

Rohit Patel, the former head of Meta's Superintelligence AI labs, left his role after eight years developing AI research to announce the founding of a new startup called Ndoc. The platform aims to provide a fully integrated AI-built alternative to replace Microsoft Office and Google Workspace, marking the latest wave of departures among senior Meta leaders following the resignations of Global Affairs head Joel Kaplan, AI researcher Andrew Toluk, and AI product chief Emily Dalton Smith.

Ndoc focuses on ending the fragmentation that modern offices experience across Excel files, Word documents, PowerPoint presentations, analytics tools such as Tableau, coding platforms and copied AI-generated text, consolidating everything into a unified workspace that colleagues can access and view without needing a registered account. The platform’s interface is organized into three main tabs: the presentation tab for outputs, slides, dashboards, simulations and micro-apps; the narrative tab, which includes deep analyses, numbered citations and explanatory figures; and the model tab, which holds the original datasets and the Python code that generates the results.

This division mirrors Patel’s career at Meta, where he focused on reinforcement learning and training intelligent agents through trial, reward and testing output reliability.The platform bets on making the full source code of documents available so users can verify every calculation without any details hidden behind the curtain.This allows a non-programmer employee to ask the chat-bot to explain the purpose of each line and confirm the equation’s correctness. The platform also lets users upload their data to generate an analytical narrative, embedding a custom context that defines the target audience and tone, relying on an extended memory that retains the context of the user, document, team and the organization as a whole.

In contrast to tools that rely on expensive commercial models, Ndoc was built and trained using a mix of open-source models, reducing its operating cost compared with systems based on closed foundational models. However, its founder’s ambition goes beyond software to reinvent the file format itself, as the platform can export documents as a single file with an open extension that any web browser can open. The desire to turn this format into an open standard was the reason Patel left Meta, since agreeing on an open standard within large companies requires the consent of hundreds of stakeholders, and he preferred to move quickly in line with AI capabilities rather than the slow pace of corporate procedures.

Ndoc is tackling this challenge with a team of only three developers and no external funding, as the founder preferred to prove market acceptance before seeking venture capital. The move comes as major labs such as OpenAI, Anthropic and Google race to launch enterprise tools tied to traditional office suites, alongside OpenAI’s recent rollout of a “Space” area within ChatGPT. Patel likens attempts to cram AI into conventional spreadsheets to putting electric lights on oxen to plow a field instead of switching to a tractor, and therefore provided integration tools that allow easy import of existing data to smooth adoption of a system whose corporate trials began with an early-access waitlist on a pay-as-you-go model.

This shift carries strategic implications for business environments in the Gulf, Egypt and the Levant;by lowering operating costs through open models, it opens the door to reassessing the massive cloud-subscription budgets that regional companies spend on desktop-seat licensesIt also gives compliance teams and risk departments in banks and investment firms a way to audit code, reducing generative-model errors in sensitive management reports. Moreover, breaking the exclusivity of file formats gives technical staff in the region greater flexibility to build analytical tools that serve local business environments, without remaining hostage to proprietary software suites that impose escalating terms and costs on institutions.