Verify it’s working

The containers are running. Before loading data, let’s confirm the system is healthy and the LLM is connected.

Test LLM connectivity

The quickest check — ask the LLM a trivial question through TrustGraph:

tg-invoke-llm '' 'What is 2 + 2?'

You should get a response within a few seconds. If this times out or errors, check the troubleshooting section in the compose guide.

Connect to the UI

Open your browser and go to http://localhost:8888.

You’ll see a login page. Select the API Key tab and enter the bootstrap token you set earlier (the value of IAM_BOOTSTRAP_TOKEN), then click Connect.

After logging in, you should see the Workflows page showing the available workflows. The Workflows button at the top right brings you back to this page from anywhere.

Run a prompt through the UI

Let’s confirm the UI can talk to the LLM. From the Workflows page, select Prompt Management.

Find the question prompt in the list on the left and select it. On the right-hand side, change the TEST box from {} to:

{"question": "What is 2 + 2?"}

Click Run. You should see the answer appear below.

This confirms the full stack is working: browser → UI → API gateway → LLM → response.

Next

Load documents — load sample data and process it into the knowledge graph.