Use Your Knowledge

Go ahead and open the OpenShift Lightspeed chat window and then ask the following question:

You should see an answer that refers to the prohibition on autoscalers and the specific processes required to get one.

Great!

What happens if you ask the question slightly differently? Do you always get the same answer following the internal policy? You might not. Why is this?

Right now, OpenShift Lightspeed is weighting all documentation as equally important. If your question doesn’t match well enough to the internal document, the generic horizontal pod autoscaling (HPA) documentation may score higher on the internal search result, which means the internal policy document isn’t used.

We understand that this isn’t optimal, and we are working on making sure that OpenShift Lightspeed prefers your internal documentation sources.

Try asking queestions about adding GPU nodes and see if you can get a response based on the internal documentation.

Try creating your own internal "documentation", rebuilding the RAG database, re-pushing the image, and seeing how that knowledge can be added to OpenShift Lightspeed responses.

OpenShift Lightspeed internally logs the conversations and requests to the model provider. Look at the logs of the OpenShift Lightspeed API server pod when asking questions so that you can see how the documentation is included and sent with the user’s question.

Take time to explore the query filter functionality.

We love user feedback and talking to customers. If you’re interested in meeting with the OpenShift Lightspeed team, please visit the product page and fill out the form to contact us: