Your own private AI Assistant

It’s 2025 after all. You thought we’d leave you stranded with just your own brain to work with? Of course not.

Now, you can obviously google things, and you could also ChatGPT things, but these scenarios are mission critical, and we can’t afford to leak any sensitive information about our scenarios! While you may have top-security level clearance, hosted services do not.

Therefore, we’ve provisioned for you an AI Assistant, running in your private project, and backed by OUR private version of Granite 3.1! You’re welcome. But be careful: With Granite power, comes Granite responsibilities.

Accessing your Assistant

  1. From the main dashboard of OpenShift AI (https://rhods-dashboard-redhat-ods-applications.PLACEHOLDER-URL.com/) choose your private project (userX)

  2. In that project, click on the link to access the Granite Assistant:

    02 02 assistant link

  3. When prompted, authenticate.

  4. Once inside the AnythingLLM interface, create a new Workspace called Granite:

    02 02 new workspace

  5. Configure the workspace with a simpler prompt: Answer the question as accurately as you are able to

    02 02 change prompt

  6. Confirm it’s properly configured by asking it a random question:

    02 02 confirm

Intended uses

We can’t think of all the things YOU might want to use this for. But we suggest the following might be ways of leveraging the AI assistant:

  • explaining terms and concepts that may not be clear

  • help with code analysis, explanation, modifications

  • error and log messages analysis

  • support with writing and developing notes

  • brainstorming ideas, theories, and tests

Getting Support during the event

  • Reach out to the moderators if you need clarifications.