Dialogflow Source Documents/Customization - Google Cloud...
I am currently working on building a DialogFlow Agent where the datastore consists of website content.
Adjusting the Number of Data Sources
One of my current challenges is figuring out how to adjust the number of sources that Vertex AI Search queries and uses in the prompt to the LLM. As of now, the system is set to use 10 sources, but I would like to customize this number to better suit my needs.
Schema Customization for Filters
Another issue I am facing is the inability to locate my current schema to potentially utilize meta tags as filters. Instead, I have made use of Google-generated fields such as "datePublished" and "dateModified" for query operators.
Controlling Datastore Responses
Is there a way to configure the datastore response so that it does not always output a link? Perhaps by setting a threshold to determine when a link should be included in the response.
Managing Irrelevant Links
One of the main challenges I encounter is that the agent sometimes outputs irrelevant links from the sources simply because they were the first links encountered. How can I gain control over this behavior and ensure that only relevant links are provided?
