RAG Integration (Watson Discovery + watsonx.ai)

This script is a simple illustrative example of integrating Watson Discovery and watsonx.ai together.

You will need Python installed locally to run this script: https://www.python.org/downloads

Most recently tested with version 3.12.2

How to use this script

First, you must create a .env file at the project root containing all sensitive data about your instances of Watson Discovery and watsonx.ai.

Here is a table of them:

Name	Description
IBM_DISC_API_KEY	API key with access to Discovery instance. Can be created here
IBM_DISC_PROJ_ID	Project ID of Discovery instance. Available within cloud console.
IBM_DISC_SERVICE_URL	Service URL of Discovery instance. Available within cloud console.
IBM_DISC_COLL_ID	Collection ID of the collection of documents you wish to query as context within Discovery.
IBM_WX_API_KEY	API key with access to watsonx.ai project. Can be created here
IBM_WX_PROJ_ID	Project ID of watsonx.ai project. Availabe under "Settings" within project.
IBM_WX_SERVICE_URL	Service URL of watsonx.ai project. Tied to locale of projects deployment.

There is a blank version in env_example. Feel free to rename this to .env and fill it with the required values. From the command line you could do:

# this command will create a copy of the file titled .env
cp env_example .env

Then you can fill out the created .env file with your instance values.

Ensure you can see hidden files within your explorer or editor to see the .env file.

Next, install the dependencies required for the script:

# installs dependencies needed for the script to run
python -m pip install -r requirements.txt

Lastly, you can now run the script via:

# runs the actual script
python rag.py

To change the question being asked simply change the QUERY variable located near the top of the script and run once more.