nao

The tool provides an open‑source framework for building analytics agents that interact with a user’s own data stack. Users define a “context” as a hierarchical collection of files—such as database schemas, column descriptions, profiling reports, and documentation—which the agent can reference when answering queries. The system includes commands to initialize this context, synchronize it from external sources like databases, Git repositories, or Notion pages, and test its reliability by running unit‑style question‑to‑SQL checks that report accuracy and performance metrics.

A chat interface lets anyone query the data in plain English, leveraging a configurable large‑language‑model key for inference. The UI can be self‑hosted, and the agent can be deployed to collaboration platforms such as Slack, Teams, WhatsApp, or Telegram, allowing multiple users to interact with the same analytics context.

The project targets developers and data engineers who need a programmable, reproducible way to expose data analytics through natural‑language interfaces while maintaining control over the underlying context and model usage. It runs on macOS, is marked as stable, and is fully open source.