What is nao?

nao is an open-source, AI-powered data IDE designed for analysts, engineers, and scientists who work with SQL, Python, or dbt workflows. It connects directly to your data warehouse, understands your schema, and helps you build, preview, and deploy data pipelines with confidence. Think of nao as an AI teammate that catches issues early, reduces bugs, and keeps your data trustworthy — all without forcing you to switch tools or contexts.

Who it's for

• Data analysts who need to write SQL queries faster, preview results, and collaborate without leaving their IDE.
• Data engineers building and maintaining dbt workflows, data pipelines, and quality checks with automated context awareness.
• Data scientists who want to run analytics, explore schemas, and deploy models while keeping their work reproducible and team-friendly.

Key features

Context engineering like a file system

nao lets you build your agent's context as a structured file system. You can add anything — data, metadata, rules, docs, tools, and MCPs — with no limit. Run nao init to create the context, then organize it however your team needs.

Synchronize with your existing stack

Pull context automatically from databases like Postgres, Snowflake, BigQuery, Databricks, DuckDB, MotherDuck, and Redshift. It also integrates with repositories (dbt, Looker, Cube, Airflow, GitHub) and external sources like Notion, Atlassian, Google Drive, and Linear — so your agent always has the latest context.

Measure and improve context reliability

Run nao test to create unit tests that turn questions into SQL. You get instant metrics on context reliability, answer rate, average response time, and total token usage. This helps you monitor performance and continuously improve your agent's accuracy.

Deploy a chat UI for your whole team

With nao chat, you can deploy a chat interface that lets anyone ask questions in plain English. The UI supports data stories, chat replay for monitoring, and works with your own LLM key (Claude, Gemini, GPT, Mistral) — so you only pay for token consumption.

What stands out

"Agent reliability depends on context. Engineer it."

Most AI tools treat context as a black box. nao flips that by making context engineering explicit and measurable. You don't just feed data to an AI — you build, sync, test, and refine the context like a software project. This means your agent actually understands your warehouse schema, business definitions, and team conventions, leading to fewer hallucinations and more trustworthy results.

Worth checking out if…

You're tired of AI tools that guess your schema or produce unreliable SQL. nao is worth exploring if your team works with modern data stacks (dbt, BigQuery, Snowflake, Databricks) and wants an open-source, self-hosted solution that gives you full control over context, costs, and deployment. It's especially valuable if you need to share analytics capabilities across non-technical team members without sacrificing data quality or security.