Chat Agent -- Retrieving information across multiple agents or dashboards that cover related topics with different variable names

Q&A
1

I am posting this question on behalf of a customer:

“About the General Knowledge agent: has anyone tested how well it works when trying to retrieve information across multiple agents or dashboards that cover related topics but use different variable names? Would the recommended approach be to provide a reference document that maps and relates the indicators across these panels for the general agent to use effectively?”

2

I asked this question to the Quick Suite Chat Agent and got this reply:

"You’re correct in identifying a key challenge: when information spans multiple agents or dashboards in Quick Suite, different variable names for the same concepts can create retrieval difficulties. Looking at the available dashboards, I can see examples of this - for instance, community metrics might be called “questions solved” in one dashboard and “solutions marked” in another.

Recommended Approach

Yes, providing a reference document that maps and relates indicators across panels would be an effective approach. Here’s why this strategy works well in Quick Suite:

1. Knowledge Base Integration

  • Create a comprehensive mapping document and add it to a Space that your General Knowledge agent can access
  • This document should include:
    • Variable name mappings (e.g., “Active Users” = “Signed-in Users” = “Community Members”)
    • Metric definitions and relationships
    • Dashboard-specific terminology guides
    • Cross-reference tables for related indicators

2. Structured Mapping Format

Your reference document should include:

  • Concept mapping: Group related metrics under common themes
  • Dashboard cross-references: Show which dashboards contain which versions of metrics
  • Calculation differences: Note if the same concept is calculated differently across sources
  • Time period alignments: Document different reporting periods used

3. Implementation Best Practices

  • Use consistent naming conventions in your mapping document
  • Include context about data sources - some dashboards may pull from different underlying datasets
  • Document semantic relationships between related but not identical metrics
  • Create a glossary of terms used across different systems

4. Quick Suite Specific Benefits

  • The General Knowledge agent can reference this mapping when users ask cross-dashboard questions
  • Spaces allow you to organize related mapping documents together
  • You can update the mapping document as new dashboards or metrics are added
  • The agent can provide more accurate responses by understanding these relationships

Alternative Approaches

If a reference document isn’t sufficient, consider:

  • Standardizing variable names across new dashboards where possible
  • Creating unified views that combine related metrics from multiple sources
  • Using Quick Suite’s integration capabilities to create consolidated data sources

This approach aligns with established best practices for knowledge management systems and should significantly improve the General Knowledge agent’s ability to provide accurate, comprehensive responses across your various dashboards and data sources."

@royyung Could you validate this ^ answer?

1 Like
3

Hi @Kristin,
I would agree with this practice being a sufficient way to handle this type of scenario. Additionally, if you’d prefer not to include additional assets, you can add a note to your chat agent that specifies the various naming conventions. Chat agents keep track of your 30 day prior history, at least for that timeframe the chat will remember your previous notes that have been added so you should not have to continuously include that prompt.
I ran some tests and was able to successfully receive correct answers, so I would suggest testing in your environment to see how the chat handles your scenario as it may differ from my testing examples!

1 Like