DB GPT

This project is an open source agentic AI data assistant. It connects to a team's databases and files, writes SQL and code autonomously, runs reusable skills in sandboxed environments, and turns the resulting analysis into reports, insights and actions. It can be deployed privately so that data and models stay under the operator's control. It is aimed at data engineers, analysts and developers building AI applications on top of their own data. Core concepts include autonomous agents for complex data tasks, AWEL (the Agentic Workflow Expression Language) for orchestrating multi step workflows, retrieval augmented generation for knowledge enhanced answers, and SMMF, a service oriented multi model management framework for running and switching between language models. Core components cover skill packages for domain specific tasks, data source connectors for databases, warehouses and data platforms, secure sandboxes for code execution, and knowledge base management. Built in applications include chat with data through natural language queries, AI analysis of CSV files, Excel workbooks and databases, and a visual builder for agentic workflows. The software is free and open source, with no paid plans on the documentation site. It can be installed with a one line installer, run from the CLI, built from source for full control, or deployed with Docker. Operating costs depend on the infrastructure and model providers chosen. The project is community developed, with its code on GitHub and models and resources on Hugging Face, and support offered through Discord and community channels rather than a commercial support contract.