Setting Up OpenWebUI with Custom Backend

This guide explains how to configure OpenWebUI to use your custom AI service as its backend.

Overview

OpenWebUI is designed to work with Ollama directly, but it can be configured to use a custom backend service instead. This allows you to add your own business logic, authentication, and other features while still using OpenWebUI's user-friendly interface.

Prerequisites

OpenWebUI installed and running (at http://104.225.217.215:8080)
Your AI service deployed and running (at http://your-server-ip:5252)

Configuration Steps

1. Locate OpenWebUI's Environment Configuration

If you're running OpenWebUI using Docker, you can modify the environment variables in your docker-compose.yml file or when running the container:

version: '3'
services:
  openwebui:
    image: openwebui/openwebui:latest
    environment:
      - OLLAMA_API_BASE_URL=http://your-server-ip:5252/ollama
      - OPENAI_API_BASE_URL=http://your-server-ip:5252/api
    ports:
      - "8080:8080"

If you're running OpenWebUI directly, you can modify the .env file in the OpenWebUI installation directory.

2. Set the API Base URLs

Add or modify the following environment variables:

OLLAMA_API_BASE_URL=http://your-server-ip:5252/ollama
OPENAI_API_BASE_URL=http://your-server-ip:5252/api

Replace your-server-ip with the IP address or hostname of the server where your AI service is running.

3. Restart OpenWebUI

After updating the environment variables, restart the OpenWebUI service to apply the changes.

If you're using Docker:

docker restart openwebui

If you're running OpenWebUI directly:

# Stop the current process
pkill -f "openwebui"

# Start OpenWebUI again
cd /path/to/openwebui
npm start

Testing the Integration

To test if the integration is working correctly:

Open OpenWebUI in your browser (http://104.225.217.215:8080)
Try to create a new chat
Select one of the models (gemma3, llama3.3, llama3.1, mistral, or deepseek)
Send a message and check if you get a response

If everything is configured correctly, OpenWebUI will send requests to your AI service, which will then forward them to the Ollama API.

Troubleshooting

If you encounter issues with the integration:

Check the logs of your AI service:
```
tail -f logs/ai_service.log
```
Check the logs of OpenWebUI:
```
docker logs openwebui
```
Verify that your AI service is running and accessible:
```
curl http://your-server-ip:5252/health
```
Verify that the OpenWebUI environment variables are set correctly:
```
docker exec openwebui env | grep URL
```

Advanced Configuration

For advanced configuration options, refer to the OpenWebUI documentation: https://docs.openwebui.com/getting-started/env-configuration

2.8 KiB Raw Permalink Blame History