Enterprise Knowledge Chat Agent - Managed AI Models as OCI Accelerator Pack

We have already described Oracle OCI AI Acceleration packs and also cookbook how to deploy one of them , Vehicle Route Optimizer. In this blog we will describe an Enterprise Knowledge Chat Agent - LLM-powered search and answer experience is grounded in your enterprise knowledge base. Choose self-hosted or managed deployment models to match sovereignty, performance, and operational preferences.

The deployment process is almost identical to the one from previous blog, we just describe the difference. For the OCI GenAI Service Region we have chosen eu-frankfurt-1, chicago did not work for us.

When you navigate to Developer Services -> Stacks -> Application Information you will see all the needed URL for use and management of the agent chat.

Take the Service URL and open the chat.

You can create a new collection, choose embedding model ,dimensions, create custom field.

After you can download files [pdf, docx, txt, mdm] which you can vectorize and use for grounding and search of your questions. The vectorizing happends against Oracle AI Autonomous Database 26ai.

After you can ask question which will be answered.

Let us see answer for question : Whom to report bribery in the public sector project ?