Report Abuse

Basic Information

DB-GPT is an open source AI-native data application development framework designed to help developers and enterprises build data-driven applications by combining large language models and databases. The project provides infrastructure for multi-model management, Text2SQL optimization, retrieval-augmented generation, generative business intelligence, automated fine-tuning, and a data-driven multi-agents framework. It includes AWEL, an Agentic Workflow Expression Language for orchestrating agent workflows, and supports connecting diverse data sources and a Data Factory for cleaning and preparing knowledge. The repository targets teams that want to create private-domain question-answering, business intelligence, and automated agents that operate over structured and unstructured enterprise data with reduced coding effort.

Links

Categorization

App Details

Features
DB-GPT bundles several core capabilities oriented to AI-native data apps. It implements a RAG framework for knowledge-based applications and unified vector storage and retrieval for large information volumes. It provides Generative BI for natural language interaction with Excel, databases and warehouses. A Text2SQL fine-tuning pipeline, including LoRA/QLoRA and SFT, is included and the README reports a Spider-dataset accuracy milestone. The project contains a data-driven multi-agents framework and AWEL for orchestration. Model management (SMMF) supports many open-source and API models. Plugin and Auto-GPT plugin compatibility, multiple datasource connectors, privacy and proxy desensitization features, and submodules for hub, visuals and app templates are also provided.
Use Cases
DB-GPT reduces the engineering work required to build bespoke, data-centric AI applications by providing ready infrastructure and workflows. Developers can use its RAG and vector retrieval to build private-domain Q&A systems. The fine-tuning and Text2SQL tooling speeds creation of SQL-capable LLMs for BI and analytics. AWEL and the multi-agent framework enable orchestrated agent behaviors and automated pipelines over enterprise data. SMMF lets teams integrate different LLMs and deploy privatized models while plugins and connectors let apps tap many data sources. Documentation, submodules and templates help adoption, and privacy and security mechanisms support enterprise deployments.

Please fill the required fields*