Meta's Rohit Patel launches Ndoc to compete with Microsoft Office and Google Workspace
Former Meta AI director launches AI-native workspace startup positioning integrated analysis, narratives, and datasets as alternative to fragmented productivity tools.
Last verified:
According to TechCrunch AI, Rohit Patel, director at Meta Superintelligence Labs, has departed Meta after eight years to launch Ndoc, a startup positioning itself as an AI-native replacement for productivity suites like Microsoft Office and Google Workspace. The departure adds to a pattern of high-profile exits from Meta’s AI organization, following the departures of Chief Global Affairs Officer Joel Kaplan, AI researcher Andrew Tulloch, and Emily Dalton Smith, head of product for AI-for-Work.
Ndoc’s three-layer architecture
Ndoc’s core design consolidates three workflows into a unified interface: the Stage tab hosts presentation outputs such as slide decks, dashboards, and simulations; the Narrative tab contains written analyses with numbered citations and generated summaries explaining findings; and the Model tab stores underlying datasets and Python scripts. This architecture reflects Patel’s conviction that AI-assisted decision-making requires auditability—users can inspect the code and formulas powering analyses rather than accepting opaque outputs.
According to TechCrunch, Patel described the user experience as enabling teams to upload datasets, request AI-driven analysis, and receive both outputs and explanations in one unified document, with optional context settings for audience, tone, and organizational memory across individuals, documents, teams, and company contexts.
Technical foundation and cost implications
Patel told TechCrunch that Ndoc uses a mixture of open-source models and was built from scratch, resulting in lower operational costs than platforms built on frontier AI models. However, Patel acknowledged that early users prioritize ease-of-use and integrated workflows over cost savings—a signal that the productivity market values consolidation over model efficiency alone.
File format and open-source ambition
The platform exports documents as “.ndoc.html” files compatible with any browser. Patel intends to open-source this file format to allow third-party developers to build applications that work with Ndoc documents without vendor lock-in, a structural choice positioning Ndoc as a platform rather than a closed ecosystem.
Why This Matters
Patel’s departure underscores growing attrition among Meta’s AI leadership and signals investor confidence in AI-first productivity tools. The startup directly challenges Microsoft and Google’s decades-long dominance in workplace software by bundling AI transparency and unified workflows as native features rather than add-ons. For teams making decisions on datasets—finance, analytics, and research—the ability to audit AI-generated analysis could become a competitive requirement rather than a feature, potentially reshaping how productivity platforms are evaluated. If Ndoc’s focus on auditability gains traction with enterprises, it may pressure incumbents to decouple AI decision-making from black-box outputs or risk losing customers to vendors prioritizing explainability.
Frequently Asked Questions
What problem does Ndoc solve?
Ndoc consolidates Excel, Word, PowerPoint, and data analysis tools into a single workspace where AI handles analysis and narrative generation while maintaining full transparency into the underlying formulas and scripts.
How does Ndoc differ from adding AI to existing tools?
Rather than layering AI onto Microsoft Office or Google Workspace, Ndoc is built from scratch as an AI-native platform with three integrated components—Stage (presentations), Narrative (analysis), and Model (datasets/code)—designed to work together.
Why is transparency important to Ndoc's design?
Patel's background in reinforcement learning and evaluating AI trustworthiness informed Ndoc's architecture; users can inspect the Python scripts and formulas behind AI-generated outputs to verify accuracy before making business decisions.