I’m trying to understand how the NotebookLM data source in Google Cloud AI Applications is intended to work with Conversational Agents / Dialogflow CX.
My end goal is to build a public-facing chat application using a CX agent, with the agent grounded in content maintained inside a specific NotebookLM notebook.
When creating a data store, I can see a NotebookLM option described as “Search through your notebooks in NotebookLM.” I expected the setup to allow me to connect the CX agent to a particular notebook, possibly by providing its notebook GUID or URL, for example:
https://notebooklm.cloud.google.com/global/notebook/<NOTEBOOK_GUID>?project=<PROJECT_ID>
Is the intended workflow to:
- Create and maintain content in one NotebookLM notebook.
- Connect that notebook to a CX agent.
- Use the CX agent for a public chat application.
- Have the agent retrieve grounded answers from the notebook sources.
- The admin should be able to add more documents to the notebook.
Or is the NotebookLM option not designed for CX agents at all?
I also enabled the 14-day NotebookLM trial, but this did not appear to unlock or change anything in the CX data-store setup. This makes me wonder whether I am mixing up two separate products or integration paths.
Could someone clarify:
- Whether NotebookLM can currently be used as a grounding source for a CX agent.
- Whether the integration requires NotebookLM Enterprise rather than the standard or trial version.
- Whether the NotebookLM option shown in AI Applications is actually intended for Gemini Enterprise rather than Conversational Agents.
- Where a notebook ID or GUID is entered, if direct notebook selection is supported.
- Whether notebook updates are automatically reflected in the agent.
- Whether there is a complete setup guide or working example for this architecture.