SQLBot is an open-source, LLM-powered text-to-SQL and conversational analytics system from the DataEase team. Business users ask questions in plain language; SQLBot turns them into SQL using retrieval-augmented generation over your schema, runs the query and returns tables and charts. The project has about 6,900 stars on GitHub and is actively developed in 2026.
Generic LLMs write plausible but wrong SQL because they do not know your tables. SQLBot grounds generation in RAG: it indexes your database metadata, terminology, custom SQL examples and data dictionaries, retrieves the relevant context for each question, and constrains the model to produce executable queries against your actual schema. It connects to mainstream relational databases and data warehouses, shows the generated SQL for transparency, visualizes results as charts, and supports follow-up questions that refine the previous query. Administrators manage data sources, permissions and row-level access, and workspaces keep teams separated. SQLBot can be embedded into other applications and exposes an MCP server interface so AI agents can query data through it as well. It deploys with Docker in minutes and works with popular LLM providers as well as local models via compatible APIs โ a practical way to give non-technical staff self-service analytics without handing out database credentials.
SQLBot is free and open source, deployable with Docker. You only pay for the LLM API usage behind it, or run a local model at no cost.
SQLBot is for companies and data teams that want business users to explore data by asking questions in chat โ without writing SQL or waiting on analysts for every report.