db-ai-assistant

Introduction

A data query framework developed based on the tio-boot framework in Java, which currently supports open-webui.

Main Features

text-to-sql: Converts user input from natural language to SQL, executes SQL, and returns visual query results

Getting Started

Launch

EXPORT OPENAI_API_KEY=""
java -jar target/db-ai-assistant-1.0.jar

Once launched, it listens on port 80. For example, my IP address is http://192.168.3.8.

Integration with open-webui

docker run -d -p 3000:8080 \
  -v open-webui:/app/backend/data \
  -e OPENAI_API_BASE_URLS="http://192.168.3.8/openai/v1" \
  -e OPENAI_API_KEYS="contact https://github.com/litongjava" \
  --name open-webui \
  --restart always \
  ghcr.io/open-webui/open-webui:main

Set OPENAI_API_BASE_URLS to the backend IP of db-ai-assistant. The OPENAI_API_KEYS can be any string; db-ai-assistant will automatically parse the message and replace it with the specified OPENAI_API_KEY.

Sending Test Messages

Test the setup:

Message Flow

The web interface receives the input message (from open-webui).
open-webui forwards the message to the application backend (db-ai-assistant).
db-ai-assistant sends the message to the large model (OpenAI ChatGPT).
ChatGPT returns the message to db-ai-assistant, which in turn sends it back to open-webui.
open-webui displays the message.

Custom Development

Developers can customize db-ai-assistant for specific domain AI agents.

litongjava/db-ai-assistant