nao is an open-source analytics agent on GitHub that lets anyone chat with their company data in plain language. The project has earned around 1,900 stars, is developed in the open with very recent activity in October 2026, and describes itself as an open-source alternative for building and deploying AI data analysts.
The nao framework has two parts. First, the nao-core command line tool helps data teams build a rich context for the agent: warehouse schemas, metadata, modeling definitions, business rules and documentation are organized in a file-system-like project that the agent understands. Second, nao provides a chat interface that can be deployed for everyone in the company, so non-technical users can ask questions and get analytics answers generated as SQL against the warehouse. It works with major stacks including BigQuery, Snowflake, Databricks and PostgreSQL, and is model-agnostic about the underlying LLM. Teams can unit test agent behavior, version the context, and track answer quality and user feedback over time before rolling changes out.
nao is open source and free to self-host from GitHub. You deploy it on your own infrastructure with your own warehouse and LLM credentials, so there is no per-seat subscription charged by the project for the open-source framework.
nao is built for data teams and analytics engineers who want to give business users a trustworthy, governed way to chat with warehouse data, and for companies that prefer an open-source, self-hosted analytics agent over a closed SaaS tool.