External Pipeline

The External Pipeline Registration feature enables GLChat to dynamically integrate with external, OpenAI-compatible AI services without code changes.

Overview

Quick Guide - End to End

Let's create your first external pipeline in under 5 minutes.

⚠️ Disclaimer: This guide is specifically for running the service and register it to GLChat. The server is not intended for use in production.

Step 1: Download and Running Your Service

To get started, download external-pipeline-example.zip

189KB

external-pipeline-example.zip

archive

Open

After you download it, you can follow the instructions (README.md) inside the downloaded folder.

Follow until you successfully run your server

In this example I set env to:

• SERVER_NAME = RAG Pipeline Attachment Support Example

• SERVER_PORT = 8006

• SERVER_NAME = Example Guide RAG External Pipeline

• SERVER_VERSION = 1.0.0

• SERVER_DESCRIPTION = Complete RAG pipeline example for GLChat external pipeline integration

• MODEL_NAMES = rag-pipeline,gpt-4o-mini

• SUPPORTS_ATTACHMENTS = true

• SUPPORTED_FILE_TYPES = '{"document": [".pdf", ".txt", ".md", ".docx"], "image": [".jpg", ".png", ".jpeg"], "audio": [".mp3", ".wav"]}'

• MAX_FILE_SIZE = 10485760

• MAX_FILE_COUNT = 5

Step 2: Register Your Pipeline in GLChat

This step will guide you to register your pipeline in GLChat

1. Open GLChat

2. Open Admin Dashboard

3. Go to RAGO Pipelines inside Pipeline & Preset

   

4. Click Register External Pipeline

   

5. Insert your url and api key service (from the step 1) and click save

   

6. Then your pipeline service is created here

   

Step 3: Interact With Your Service

To interact with your service, you need to create your application and use your pipeline service as your RAG Pipeline.

1. Create your application (new application) from Application page

   

2. Fill required fields

3. The one you need to change is RAG Pipeline field. Set it to your newly created pipeline service

   

4. Click next and save until your application created

5. Go back to your Application

6. You can now select your newly created application

   

7. If you follow the step 1, the service can answer your question with the context from knowledge base

8. Try chat yourself!

   

9. If you using the external pipeline server from Step 1, it can also handle Attachment and Web Search

   

Next section will have more detailed on how you can implement your external pipeline service.