vanna-ai

vanna: Turn Natural-Language Questions into SQL Insights

Vanna helps teams build chat interfaces that translate natural-language questions into SQL and return streamed tables, charts, and summaries. It is aimed at applications that need database-backed answers with user-aware permissions and an embedded web UI.

PythonMITAILLMDatabase
View on GitHub Compare
Stars
23.8k
Forks
2.5k
Stars, last 30 days
-19

Our take

Health score

34/100

How it is scored
Activity
36
Community
60
Issues
21
Pull requests
20

-19 stars in the last 30 days

Project history

Daily GitHub snapshots since Oct 7, 2026. History covers stars and forks; health signals are not tracked historically.

Selected snapshot

Stars

23,798

Forks

2,547

Change since Nov 16, 2026: -5 stars·0 forks

Overview

Vanna is a Python framework for building conversational interfaces to SQL databases. It connects an LLM-driven agent to database tools so users can ask questions in natural language and receive query-backed answers, rather than writing SQL themselves.

Its newer 2.0 direction is focused on production applications: it includes a web chat component, streaming responses, and mechanisms for passing user identity and permissions through the request. It may suit teams embedding analytics in an existing app, while simpler one-off query workflows may not need its agent and server setup.

Key Features

  • Translates natural-language questions into SQL through an LLM-backed agent and database tools.
  • Streams progress and response components, including tables, charts, and summaries.
  • Provides an embeddable <vanna-chat> web component for React, Vue, or plain HTML pages.
  • Supports user identity resolution and permission-aware tool execution, including row-level filtering described in the README.
  • Integrates with FastAPI and Flask and allows custom tools, lifecycle hooks, and LLM middleware.
  • Documents integrations for multiple LLM providers and database systems.
  • Includes tracing and metrics integration for observability.

Use Cases

  • Product teams adding a natural-language analytics interface to an application backed by SQL data.
  • SaaS developers who need database answers to respect user or group permissions.
  • Data teams that want to let non-SQL users explore data through a chat UI.
  • Developers integrating streamed tables and charts into an existing web app and authentication flow.

Getting Started

Start with the Quickstart, then see the configuration guide and README for setup details. The README does not specify a single install command.

Considerations

  • The repository is marked archived, so check its current status and support expectations before adopting it for a new production system.
  • A working deployment needs an LLM service, a database connection, and application-side configuration; user authentication and permission rules also need to be integrated for protected data.
  • The README describes Vanna 2.0 as a complete rewrite from 0.x, with a different Agent API. Existing users should review the migration guide before upgrading.
  • The web component and server provide building blocks, but teams remain responsible for configuring their own data access and security policies.

Found this useful?

Share it with someone who would like vanna.

Comparisons