OpenChatBI enables users to query and analyze databases through natural language conversations by leveraging LLMs to generate SQL queries from text inputs. Built on LangGraph and LangChain, it orchestrates multi-step agent workflows that include schema linking, text2sql conversion, code execution for analysis, and time series forecasting. The system maintains persistent conversation memory, supports knowledge base retrieval via MCP tools, and provides both simple and streaming web interfaces built with Gradio and Streamlit.