# nao (full)

> nao is the open source agentic BI builder and analytics agent. Data teams sign up on nao Cloud (app.getnao.io) or self-host, connect their warehouse, and build and control the context the agent runs on (schemas, metrics, rules, docs) as files in Git, test it with nao test, and improve it from real usage. Everyone in the company can then chat with their data from the nao app, Slack, Teams, WhatsApp, Telegram or Claude and ChatGPT through MCP, and turn answers into live stories (dashboards), data apps and slides in their own design.

## Canonical facts
- Product name: nao (always lowercase)
- Company: nao Labs, founded 2024, Y Combinator (Spring 2025), offices in Paris and San Francisco
- Founders: Claire Gouze (CEO) and Christophe Blefari (CTO)
- License: Apache 2.0 open source core, self-hostable, or hosted on nao Cloud
- Install: `pip install nao-core` (CLI), Docker image `getnao/nao`
- Website: https://getnao.io/
- GitHub: https://github.com/getnao/nao
- Documentation: https://docs.getnao.io/
- Trust Center: https://compliance.getnao.io/
- Contact: https://getnao.io/contact/

## What nao does
- Chat with your data everywhere: the nao web app, Slack, Microsoft Teams, WhatsApp, Telegram, and Claude or ChatGPT through the nao MCP server. Every conversation is visible to the data team in chat replay.
- Stories: any answer becomes a live dashboard with filters, tabs, KPIs, charts and tables. Stories refresh on a schedule, can be shared, pinned and certified, and people can chat with them.
- Data apps: custom stories add inputs and logic (planners, simulators, internal tools).
- Slides: present any story as a slide deck.
- Custom design: your fonts, colors and logo on every story and slide.
- Automations: run a prompt on a schedule (hourly, daily, weekly, monthly) and send the answer to Slack, email or GitHub.

## How data teams control it
- Context as files: `nao sync` writes one folder per table (columns, preview, profiling, description) into your Git repo, next to RULES.md, docs, dbt, semantic layer YAML and skills. The agent searches these files directly, there is no opaque RAG.
- Evaluation: `nao test` runs question-to-SQL unit tests and compares the data returned (a data diff, not an LLM judge) to give a reliability score before each deployment and in CI.
- Context recommendations: nao reads real conversations, errors and feedback and proposes context fixes as pull requests.
- Bring your own LLM key from any provider, with no margin on tokens and budgets per project.

## Plans
- Open source: free, self-hosted, unlimited users. Same product core: context, chat UI, stories, tests, chat replay, Slack, Teams, MCP.
- nao Cloud: free up to 3 users, then $50 per user per month, or $2K per month for unlimited users. Hosted in Europe, no annual commitment, priority support.
- nao Enterprise: annual license per seat with volume discounts and a cap, viewers free. Adds SSO with Microsoft Entra ID, Okta and OIDC, group mapping, multiple projects, row-level security, warehouse identity passthrough, white label, SOC 2 Type II reports and priority support. Same deployment as open source, activated with a license key.
- Pricing page: https://getnao.io/pricing/
- Enterprise page: https://getnao.io/enterprise/

## Security and data
- Self-hosted deployments run on your infrastructure (Docker and Postgres on GCP, AWS, Azure or Kubernetes). nao never sees your data; only a license ping is sent, and an offline license exists.
- nao Cloud is hosted in Europe on GCP. Data stays in your warehouse; only encrypted credentials are stored.
- SOC 2 Type II. The open-source code is fully auditable.

## FAQ
### If nao is open source, what do I pay for?
The core product is open source: context, chat UI, stories, tests, chat replay, Slack, Teams and MCP. nao Cloud is the same product, hosted and supported by us.

nao Enterprise adds governance and security: SSO, your warehouse permissions in nao, multiple projects, white label and SOC 2 reports. If everyone can see the same data, open source is enough; if not, go Enterprise.

### How much does nao cost?
Open source is free, and nao Cloud is free up to 3 users, then $50 per user per month, capped at $2K per month: from 40 users, users are unlimited.

Enterprise is an annual per-seat license, with volume discounts, a cap, and free viewers. In every plan you bring your own LLM key, and nao takes no margin on tokens.

### How do I know the numbers nao gives are right?
nao test runs your key questions against the SQL you expect and compares the data, so you get a reliability score before every deployment.

Once live, chat replay shows every conversation and context recommendations suggest fixes as pull requests. Most teams reach 80 to 90% on their tests before opening nao to users.

### Does nao respect our data access rights?
In open source, everyone in the project sees the same data.

With Enterprise, nao applies each user's own warehouse permissions through your SSO, so the agent only sees what that person can see. You can also run multiple projects to keep domains like HR and finance fully separate.

### Why nao rather than Claude with an MCP server, or building our own agent?
With Claude and a warehouse MCP, there is no shared context, no visibility for the data team and no per-user permissions. Building in-house usually takes at least two full-time people.

With nao, you own the context layer as files in Git, and nao gives you the agent, the UI and the evaluation.

### Which LLMs does nao support, and who pays for tokens?
Any provider: Anthropic, OpenAI, Gemini, Bedrock, Azure OpenAI, Mistral, OpenRouter, Ollama and open models. You bring your own key, so you pay your provider directly.

Only a sample of rows reaches the model, and you can set budgets and see the cost of every conversation.

### How does the context work? Do I need a semantic layer?
The context is a set of files in your Git repo: one folder per table, plus RULES.md, dbt, docs and skills. The agent searches these files directly, with no opaque RAG.

You don't need a semantic layer to start. If you have one, add its YAML as context.

### How long does it take to get started?
A first context runs locally in about an hour, and a solid use case takes about a week.

Start small, with one domain and 10 to 20 well documented tables, then grow. You need one data person to own the context.

### Where does our data go?
Self-hosted, nao runs on your infrastructure and we never see your data; only a license ping is sent. On nao Cloud, data stays in your warehouse and only encrypted credentials are stored.

nao is SOC 2 Type II, and the code is fully auditable.

### Can nao replace our BI tool?
Any answer can become a story: a live dashboard that refreshes, can be shared and certified, and that people can chat with. Custom stories go further and work as data apps.

Many teams keep their BI tool for complex dashboards at first and move day-to-day reporting to nao.

### Where can people chat with nao?
In the nao web app, Slack, Microsoft Teams, WhatsApp, Telegram, and Claude or ChatGPT through the nao MCP server. All of it is open source, and every conversation shows up in chat replay.

### How do we try nao or run a proof of concept?
Install nao-core and test it locally, or start on nao Cloud, free up to 3 users.

For larger companies, we run a one-month proof of concept on Enterprise and help with the setup. Going to production keeps the same deployment and context.

## Key pages
- Home: https://getnao.io/
- Pricing: https://getnao.io/pricing/
- Enterprise: https://getnao.io/enterprise/
- FAQ: https://getnao.io/product/faq/
- User stories: https://getnao.io/users/
- Compare nao with other tools: https://getnao.io/resources/compare/
- Agentic Analytics Playbook: https://getnao.io/resources/playbook/
- Events: https://getnao.io/resources/events/
- Blog: https://getnao.io/blog/
- About: https://getnao.io/company/

## Documentation
- Getting started: https://docs.getnao.io/
- Context engineering playbook: https://docs.getnao.io/nao-agent/context-engineering/playbook
- Self-hosting: https://docs.getnao.io/nao-agent/self-hosting/overview
- Enterprise: https://docs.getnao.io/nao-agent/enterprise/overview

## Integrations
- [Anthropic](https://getnao.io/product/integrations/anthropic/): Claude AI models for advanced reasoning and analysis
- [OpenAI](https://getnao.io/product/integrations/openai/): GPT models for natural language processing and generation
- [Gemini](https://getnao.io/product/integrations/gemini/): Google's multimodal AI models for diverse tasks
- [Amazon Bedrock](https://getnao.io/product/integrations/bedrock/): Fully managed service that makes foundation models from leading AI startups and Amazon available via an API.
- [dbt Core](https://getnao.io/product/integrations/dbt-core/): Transform data in your warehouse with SQL-based workflows
- [BigQuery](https://getnao.io/product/integrations/bigquery/): Google Cloud's serverless, scalable data warehouse
- [Snowflake](https://getnao.io/product/integrations/snowflake/): Cloud data platform for data warehousing and analytics
- [Postgres](https://getnao.io/product/integrations/postgres/): Open-source relational database management system
- [Databricks](https://getnao.io/product/integrations/databricks/): Unified analytics platform and lakehouse for data engineering, science, and AI
- [DuckDB](https://getnao.io/product/integrations/duckdb/): In-process SQL OLAP database management system, ideal for analytics workloads.
- [Amazon Athena](https://getnao.io/product/integrations/athena/): Serverless, interactive query service to analyze data in Amazon S3 using standard SQL.
- [Amazon Redshift](https://getnao.io/product/integrations/redshift/): Fully-managed data warehouse service in the cloud for analytics and BI.
- [MotherDuck](https://getnao.io/product/integrations/motherduck/): Serverless analytic database powered by DuckDB, optimized for the cloud.
- [MCPs](https://getnao.io/product/integrations/mcp/): We support all custom MCPs. Bring whatever you need to nao agent.
- [Airbyte](https://getnao.io/product/integrations/airbyte/): Open-source data integration platform for ELT pipelines, syncing data from sources to destinations.
- [dlt](https://getnao.io/product/integrations/dlt/): An open-source Python library to build data pipelines and load data into destinations like warehouses and lakes.
- [Omni](https://getnao.io/product/integrations/omni/): All-in-one business intelligence platform for analytics, dashboards, and data collaboration.
- [Metabase](https://getnao.io/product/integrations/metabase/): Open-source business intelligence tool for exploring and visualizing data from your database.
- [Elementary](https://getnao.io/product/integrations/elementary/): Open-source data observability and reliability tool for modern data stacks, including dbt and more.
- [Select Start](https://getnao.io/product/integrations/select-start/): Metadata Context Platform for Data & AI
- [Tableau](https://getnao.io/product/integrations/tableau/): Leading analytics and business intelligence platform for interactive data visualization.
- [Notion](https://getnao.io/product/integrations/notion/): All-in-one workspace and productivity platform for notes, tasks, wikis, and databases.
- [Obsidian](https://getnao.io/product/integrations/obsidian/): Knowledge base that works on top of a local folder of Markdown files.
- [GitHub](https://getnao.io/product/integrations/github/): Code hosting platform for version control and collaboration, with powerful developer APIs.

## Compare pages
- [nao vs Claude + MCP](https://getnao.io/resources/compare/claude-mcp/): Anthropic's model with Model Context Protocol tools
- [nao vs Claude + Plugins](https://getnao.io/resources/compare/claude-plugins/): Anthropic's Claude with custom plugins for data access
- [nao vs Claude Dashboards](https://getnao.io/resources/compare/claude-dashboards/): Anthropic's prompt-to-dashboard feature inside Claude
- [nao vs Dust](https://getnao.io/resources/compare/dust/): AI assistant platform for knowledge workers
- [nao vs Dagster Compass](https://getnao.io/resources/compare/dagster-compass/): Slack-native AI data assistant by Dagster Labs
- [nao vs Cursor](https://getnao.io/resources/compare/cursor/): AI-first code editor for software engineers
- [nao vs OpenAI Codex](https://getnao.io/resources/compare/codex/): OpenAI's cloud-based coding agent
- [nao vs TextQL](https://getnao.io/resources/compare/textql/): Enterprise-grade text-to-SQL with data mesh integration
- [nao vs Julius](https://getnao.io/resources/compare/julius/): AI data analyst for spreadsheets and databases
- [nao vs Fabi.ai](https://getnao.io/resources/compare/fabi/): AI-powered analytics for self-serve business intelligence
- [nao vs Cube](https://getnao.io/resources/compare/cube/): Semantic layer and headless BI platform
- [nao vs ThoughtSpot](https://getnao.io/resources/compare/thoughtspot/): AI-powered analytics with natural language search
- [nao vs Snowflake Cortex](https://getnao.io/resources/compare/snowflake-cortex/): AI features built natively into Snowflake
- [nao vs Databricks Genie](https://getnao.io/resources/compare/databricks-genie/): Conversational analytics built into Databricks
- [nao vs ClickHouse + LibreChat](https://getnao.io/resources/compare/clickhouse-librechat/): Self-hosted AI chat connected to ClickHouse via MCP
- [nao vs Power BI](https://getnao.io/resources/compare/power-bi/): Microsoft's BI platform, with Copilot bolted on
- [nao vs Tableau](https://getnao.io/resources/compare/tableau/): Salesforce's visual BI platform, with almost no AI
- [nao vs Looker](https://getnao.io/resources/compare/looker/): Enterprise BI platform with LookML semantic layer
- [nao vs Metabase](https://getnao.io/resources/compare/metabase/): Open-source BI tool with AI question-answering
- [nao vs Omni](https://getnao.io/resources/compare/omni/): AI-native BI platform built on dbt
- [nao vs Lightdash](https://getnao.io/resources/compare/lightdash/): Open-source BI built on top of dbt
- [nao vs Hex](https://getnao.io/resources/compare/hex/): AI-native collaborative notebook and data app platform
- [nao vs Count](https://getnao.io/resources/compare/count/): AI-native analysis canvas for data teams
- [nao vs Agno Dash](https://getnao.io/resources/compare/agno-dash/): Self-learning open-source SQL agent
- [nao vs LangChain](https://getnao.io/resources/compare/langchain/): Open-source framework for building LLM applications
- [nao vs LibreChat](https://getnao.io/resources/compare/librechat/): Self-hosted open-source AI chat platform
- [nao vs Vercel Knowledge Agent](https://getnao.io/resources/compare/vercel-knowledge-agent/): File-system based open-source knowledge agent template

## Blog
- [Basics of Git data analysts should learn](https://getnao.io/blog/git-basics-for-data-professionals/): Learn Git essentials for data professionals, including practical workflows and how AI can help automate version control for SQL, Python, and data projects
- [How to choose the right data stack](https://getnao.io/blog/data-stack-guide/): A comprehensive guide to choosing the right data stack for your company's stage and data maturity.
- [How to Build Your In-House Analytics Agent Fully with Open Source](https://getnao.io/blog/build-open-source-analytics-agent-text-to-sql/): A practical 7-step framework to build your in-house analytics agent fully with open source tooling, from context engineering to evaluation and rollout.
- [5 Reasons Natural Language Analytics is Replacing Traditional BI Dashboards](https://getnao.io/blog/natural-language-replacing-bi-dashboards/): Why natural language interfaces are displacing traditional BI dashboards - and how to prepare your data warehouse for this shift
- [How to Build Production-Ready AI Agents for Data Analytics: The Complete Guide](https://getnao.io/blog/build-production-ready-ai-agents/): A complete guide to building reliable AI agents for data analytics - from context engineering to production deployment
- [Microsoft Entra ID SSO is GA in nao Enterprise](https://getnao.io/blog/azure-entra-id-sso-ga/): nao Enterprise now supports Microsoft Entra ID SSO in general availability, with identity-preserving runtime queries for teams using Azure AD federation.
- [5 Steps to Deploy an Analytics Agent on dbt MCP to Your Whole Company](https://getnao.io/blog/deploy-analytics-agent-dbt-mcp-steps/): A practical 5-step setup guide to deploy an analytics agent on dbt MCP with nao, from choosing the right MCP to rolling out chat with data across your company.
- [20 Best AI Analytics Agents Compared: 2026 Benchmark](https://getnao.io/blog/ai-data-agents-compared/): Full benchmark of 20 AI analytics agents on accuracy, cost, context depth, and data team UX. Warehouse-native tools, AI-native BI, and open-source agents compared.
- [Launching nao agentic BI builder](https://getnao.io/blog/launching-nao-agentic-bi-builder/): Sign up on nao Cloud, connect your warehouse, add business context, and deploy your own agentic BI to the whole company from the nao UI. No CLI needed. Still open source.
- [Launching nao Context Engineering skills](https://getnao.io/blog/launching-nao-skills/): Five open-source skills that set up your nao project, write its rules, build its test suite, audit it, and add a semantic layer. Install with `nao skills add getnao/nao`.
- [Launching nao automations](https://getnao.io/blog/launching-nao-automations/): Schedule your nao analytics agents to run in the background. Recurring reports, conditional alerts, and a new Feed to track all agent activity.
- [How Data Teams Build Skills for Agentic Analytics: 7 Practical Steps](https://getnao.io/blog/claude-skills-for-agentic-analytics/): A practical 7-step guide for data teams to design, organize, and test skills that improve analytics agent reliability and chat-with-data outcomes.
- [How to Evaluate an Analytics Agent: A Practical Guide with nao test](https://getnao.io/blog/analytics-agent-benchmark-context-stack/): A step-by-step guide to evaluating your analytics agent's reliability using nao's built-in unit test framework — from writing your first test to running the visual dashboard.
- [Launching nao Stories](https://getnao.io/blog/how-to-turn-chat-with-data-into-shareable-stories/): Stories turn a chat with your data into a shareable, editable, versioned report. Watch the launch demo and read how it works.
- [What Is the Impact of a Semantic Layer on Analytics Agent Performance?](https://getnao.io/blog/semantic-layer-impact-analytics-agent/): We tested MetricFlow semantic layers against a plain rules.md file to measure the real impact on analytics agent reliability, cost, and speed. Here's what the data says.
- [Launching the nao MCP App](https://getnao.io/blog/launching-nao-mcp-app/): nao MCP now renders interactive charts and stories directly inside your AI agent. Same nao UI, no tab switch.
- [LangChain vs Wren AI vs nao: which open-source tool for agentic analytics?](https://getnao.io/blog/langchain-vs-wren-ai-vs-nao/): Three open-source projects for agentic analytics, tested on the same BigQuery question. Setup, accuracy, context approach, and which to pick.
- [Launching nao context recommendations](https://getnao.io/blog/launching-nao-context-recommendations/): nao now audits its own usage and tells you exactly where your context is missing, wrong, or unclear - then opens the fix as a pull request in your context repo.
- [What Context Has the Most Impact on Analytics Agent Performance?](https://getnao.io/blog/context-impact-analytics-agent/): We tested schema, data sampling, profiling, dbt repos, and rules.md to find which context pieces actually improve analytics agent reliability. Here's what the data says.
- [Launching nao Enterprise](https://getnao.io/blog/launching-nao-enterprise/): nao Enterprise gives data teams SSO, row-level security, branded UI, and implementation support - while staying open source at the core.
- [How to Build a Context Stack for Agentic Analytics](https://getnao.io/blog/how-to-build-context-stack-for-agentic-analytics/): A practical 7-step guide for data teams to build a context stack that improves analytics agent reliability, speed, and cost control.
- [How to Build an AI-First Data Team](https://getnao.io/blog/ai-first-data-team-steps/): A practical 5-step guide to turning your team into an AI-first data team with context engineering, open source analytics, and chat with data workflows.
- [Open source alternative to Claude Dashboards](https://getnao.io/blog/open-source-alternative-to-claude-dashboards/): Claude Dashboards ties your BI to one agent harness. Here is the open source alternative: same dashboard, your context in Git, tested and observable.
- [How to do context engineering for analytics agents](https://getnao.io/blog/how-to-do-context-engineering-for-data-teams/): A summary of 3 context engineering studies on analytics agents: what actually moves reliability, whether a semantic layer is worth it, and how to set up a testing and monitoring framework.
- [5 Best Open Source Analytics Agents (2026 Comparison)](https://getnao.io/blog/open-source-analytics-agent-builder-playbook/): Compared 5 open source analytics agents on SQL accuracy, context depth, and production readiness. Find the right stack for your data team.
- [What's the Best Analytics Agent for Your Data Team?](https://getnao.io/blog/best-analytics-agent-for-data-team/): A practical guide to choose the best option for you to deploy an analytics agent.
- [The Agentic Analytics Playbook is out](https://getnao.io/blog/agentic-analytics-playbook/): Learn how to choose your harness, build your context layer, plan your rollout, measure success, and get examples from 7 real-life companies.
- [Why data teams need an open framework for context engineering](https://getnao.io/blog/open-framework-context-engineering/): Analytics agents won't scale without a proper context stack. Here's why context engineering needs its own open framework — and what that framework looks like.
- [Launching the nao MCP](https://getnao.io/blog/launching-nao-mcp/): nao now exposes itself as an MCP server. Run governed analytics from Claude, Cursor, Codex, or any MCP-compatible tool.
- [How to Set Up an AI Analytics Teams Bot with an Open Source Framework](https://getnao.io/blog/setup-ai-analytics-teams-bot-open-source/): A step-by-step guide to set up an AI analytics Microsoft Teams bot with an open source framework so your team can chat with data directly in Teams.
- [How to Set Up an AI Analytics Slack Bot with an Open Source Framework](https://getnao.io/blog/setup-ai-analytics-slack-bot-open-source/): A practical step-by-step guide to set up an AI analytics Slack bot with an open source framework so your team can chat with data directly in Slack.
- [Why we're making our Analytics Agent open source](https://getnao.io/blog/why-open-source-analytics-agent/): Open source isn't a marketing strategy. It's the only way to build trust in analytics agents — and the only way context engineering matures as a discipline.
- [Introducing the First Headless Analytics Agent](https://getnao.io/blog/introducing-headless-analytics-agent/): nao is becoming the first headless analytics agent - your company data brain, everywhere you work. No dashboard needed.
- [4 Steps to Improve Your Analytics Agent Reliability from 45% to 86%](https://getnao.io/blog/improve-analytics-agent-reliability-steps/): A practical case study on how context engineering, dbt documentation, and data-model fixes improved an analytics agent from 45% to 86% reliability.
- [We're launching the first Open Source Analytics Agent Builder](https://getnao.io/blog/open-source-analytics-agent-launch/): We're open sourcing nao — an analytics agent framework built on context engineering. Here's our vision for what comes after black-box BI.
- [How to Set Up dbt Core (2026): Your First Model with AI](https://getnao.io/blog/dbt-setup-guide/): Learn to set up dbt Core in 2026: install the CLI, write staging and fact models on BigQuery, add tests, and use AI to build 10x faster.
- [How to Do Data Modeling for AI Agents: 8 Practical Rules](https://getnao.io/blog/how-to-do-data-modeling-for-ai-agents/): A practical guide to data modeling for AI agents, with clear rules to optimize for precision, reliability, and better chat-with-data performance.
- [Agentic Analytics Meetup Paris 🇫🇷](https://getnao.io/blog/agentic-analytics-meetup-paris/): Recap of the first meetup dedicated to agentic analytics: 4 data teams from Gorgias, Malt, GetAround, and The Working Company share what they've actually built, what worked, and what didn't.
- [How to Make the Semantic Layer Work for Analytics Agents](https://getnao.io/blog/how-to-make-semantic-layer-work-for-analytics-agents/): How to go from the semantic layer answering no questions to 82% reliability on a wide business range - 4 steps with real benchmark data.
- [nao has a new look](https://getnao.io/blog/nao-redesign/): We rebuilt the nao interface from the ground up. New home screen, a prompt queue, visible agent reasoning, redesigned charts and stories.

## Key terms
- Analytics agent: an agent that answers business questions from your warehouse using governed context.
- Agentic BI: BI where people chat with an agent and turn answers into dashboards, data apps and slides themselves.
- Context engineering: designing, versioning, testing and improving the information an agent uses.
- Context layer: the files (schemas, metrics, rules, docs, tools) the agent reads, owned by the data team in Git.