DOCSEARCHbyAlgolia
Submit Docs

nao

website·Indexed 34 days ago

https://docs.getnao.io

An open-source framework for building and deploying analytics agents. nao allows teams to create context-aware agents that interact with data warehouses and databases (such as BigQuery, Athena, and MotherDuck) using natural language. It features a chat interface, automated reporting via Stories, and integrations with Slack, Microsoft Teams, Telegram, WhatsApp, and Mattermost. The platform includes tools for context management, MCP (Model Context Protocol) server capabilities, and a hosted version called nao Cloud.

Tokens
70.7K
Snippets
220
Records
523
Agent score
63%

What's inside nao

  1. Overview of the nao Analytics Agent Framework

    nao is an open-source framework used to build and deploy analytics agents. It enables developers to create a comprehensive context for an agent—including data, metadata, modeling, and rules—using the nao-core CLI. Once configured, it deploys a UI that allows users to perform analytics on their data via a chat interface.

  2. Overview of nao agent capabilities

    nao agent is an AI-powered assistant that has access to the codebase repository and data connection metadata. It can search the codebase and data warehouse metadata, execute queries on the warehouse, run commands, and integrate with the data stack via Model Context Protocol (MCP) tools.

  3. Overview of nao Cloud

    nao Cloud is the hosted version of nao available at app.getnao.io. It provides a managed environment including the chat application, database, and project hosting, removing the need for users to manage their own infrastructure. Users build their agent's context locally or in Git and push it to their Cloud workspace.

  4. Overview of the nao Microsoft Teams Bot

    The nao Teams bot integrates your analytics agent directly into Microsoft Teams. It allows users to ask questions and receive insights (including SQL queries, data tables, and visualizations) within Teams channels, group chats, and direct messages. The bot shares the same agent context, rules, and capabilities as the nao web chat interface; any updates to the agent are automatically reflected in Teams.

  5. Overview of nao IDE

    nao IDE is an AI-powered data editor designed for data teams. It connects directly to a user's data warehouse and business context, utilizing an AI agent to write code while maintaining data quality.

  6. Overview of User Groups and Access Logic

    User Groups are project-level, admin-only settings used to control feature access, context, conditional rules, and data row access. They complement project roles: roles define what a user can do in the platform, while groups define what features and data are available within a specific project.

    Access follows 'allow-only union' semantics: a user's effective access is the sum of all groups they belong to. If any group grants a permission, the user has that permission. There are no 'deny' rules that can override an 'allow' from another group.

    Every project contains a default All Users group that includes everyone and cannot be modified.

  7. Overview of nao agent built-in tools

    The nao agent autonomously uses a set of built-in tools to answer user requests. These tools include:

    • execute_sql: Runs SQL against connected databases. Use duckdb_local to query local CSV, JSON, Parquet, and Excel files. The save_to parameter allows writing results to permanent storage.
    • display_chart: Renders Bar (including stacked), Line, Pie charts, and KPI cards in chat or Slack.
    • execute_sandboxed_code: Runs Python and shell code in an isolated micro-VM. It can install Python packages and access images or permanent storage files. Requires enabling Sandboxes in Settings -> Agent -> Capabilities.
    • write: Saves files to the user's /home folder (the only writable directory). Unavailable if NAO_STORAGE_BACKEND=none.
    • search, list, read, grep: Tools for navigating and searching the project context files using glob patterns and regex.
    • web_search / web_fetch: Uses provider-native tools (OpenAI, Anthropic, Google) for external information. Can be enabled per project in Settings -> Agent -> Capabilities.
    • clarification: Prompts the user with a 'Quick question' card and up to 5 clickable answer chips when a request is ambiguous.

  8. Overview of the nao WhatsApp Bot

    The nao WhatsApp bot integrates your analytics agent into WhatsApp, allowing users to ask questions and receive insights (including SQL queries, data tables, and visualizations) directly in chat. It shares the same agent context, rules, capabilities, and user memory (preferences, definitions, and notes) as the web chat interface.

  9. Overview of nao Recommendations for online evaluations

    Recommendations provide 'online evaluations' by auditing nao's own production usage to identify gaps, errors, or ambiguities in project context. Unlike nao test (offline evals used in CI against a fixed benchmark), Recommendations mine real user interactions to find where context is missing, wrong, or unclear.

  10. Overview of nao Context Engineering Skills

    nao provides six published skills to automate the context-engineering lifecycle. These skills are designed to be loaded by agents (such as Claude Code, Codex, or Cursor) via SKILL.md files to ensure high-reliability data context and metric definitions.
    setup-context         →  write-context-rules   →  create-context-tests  →  audit-context (anytime)
(first time only)        (any rules change)       (benchmark + extend)     (diagnose, never fix)
                                                          │
                                                          ▼
                                                   tests reveal metric
                                                   reliability gaps?
                                                          │
                                                          ▼
                                                  add-semantic-layer
                                                  (then back to write-context-rules)

Ready to ship?
      │
      ▼
deploy-context
(CI/CD to production)

  11. Overview of the nao Telegram Bot integration

    The nao Telegram bot integrates your analytics agent into Telegram, allowing users to ask questions and receive insights (including SQL queries, data tables, and visualizations) directly within the chat interface. The bot shares the same agent context, rules, and capabilities as the nao web chat interface; any updates to the agent are automatically applied to the Telegram bot.

  12. Overview of the nao Mattermost Bot

    The nao Mattermost bot integrates your analytics agent into a self-hosted Mattermost server. It allows teams to ask questions and receive insights (including SQL queries, data tables, and visualizations) directly within their workspace. The bot uses the same agent context, rules, and models as the nao web chat interface. Each nao project connects its own bot, allowing multiple projects to coexist on one Mattermost server.