Before asking the AI Agent to build something new, you can ask it a different question first: do we already have a dashboard for this? Describe the business question or the kind of analysis you're after, and the Agent will search your existing dashboards and workbooks to see whether a strong answer already exists. If it finds one, it surfaces that content directly instead of generating something new, saving you time and putting proven, previously built analysis back in front of you.
This only works well, though, if your content is easy to find. This article explains how the Agent decides what to surface when asked, and what you can do as a dashboard builder or folder owner to make sure your work gets found.
This Is Different From the Search Bar
This is not the global Search bar at the top of Insights, and it works differently. Search is a fast, filter-based way to jump to a dashboard when you already know roughly what you're looking for, using keywords or filters like type: or in:. For a full guide to that experience, see Using Enhanced Search in Insights.
Asking the Agent is a more conversational way to discover content. Instead of a title or a filter, you describe a business question or the kind of analysis you're after, in your own words, inside the Agent's chat. Because it's a conversation, you can go back and forth: add detail, rephrase, or clarify if the first answer isn't quite what you meant.
How the Agent Finds Existing Content
When you ask the Agent whether a dashboard already covers your question, describing the business question or analysis in your own words, it searches across your dashboards and workbooks and ranks them by relevance before deciding whether to surface one or suggest building something new. That ranking is based on a combination of signals:
Dashboard name: how closely your question matches the dashboard's title
Tile titles: how closely your question matches the names of individual tiles on the dashboard
Dashboard and tile descriptions: whether the written description of the dashboard, or of a tile, covers what you're asking about
Folder name: whether the folder the dashboard lives in reflects the topic
Creator: who built the dashboard
Recency: how recently the dashboard was updated, used as a tiebreaker between otherwise similar matches
Why This Is Worth Your Attention
The details that help a teammate find the right dashboard, a clear name, a helpful description, a sensible folder, are the same details the Agent relies on. Think of the Agent as another user searching your library. If a person could find and understand your dashboard by its name and description alone, the Agent likely can too.
That means the effort you put into good dashboard hygiene pays off twice: once for the humans browsing and searching Insights, and again for the Agent, which can now direct people straight to your existing work instead of rebuilding it from scratch.
What to Do to Improve Discoverability
A few habits make the biggest difference:
Use clear, specific titles for both the dashboard and its individual tiles. A name like "Q3 Group Pace by Segment" is far easier to match than "Dashboard 2" or "Copy of Pace Analysis."
Add a description to the dashboard itself, covering what it measures, what it's for, and who it's built for. You can write your own or accept the AI-suggested description when you save a document.
Add descriptions to key tiles, not just the dashboard as a whole. Tile-level descriptions let the Agent match on what a specific chart or table actually shows, not just the dashboard's overall theme.
Keep dashboards in well-named folders. A folder name that reflects its contents adds another signal the Agent can use to connect a question to the right dashboard.
You can add or edit a description on any existing dashboard at any time by going to Edit > Rename.
The Bottom Line
If you build or maintain dashboards, treat titles, descriptions, and folder structure as part of the job, not an afterthought. It is the difference between your work being found and reused, and it being duplicated by someone (or something) that never knew it existed.
If you have any questions or want help exploring this, click the ? icon at the bottom of the left-hand menu. We're happy to help.