A Gleap custom agent can use Metabase through Composio: it searches questions and dashboards and runs saved questions with parameters. Teams that already model their data in Metabase get the same numbers in support conversations that analysts use.
What a Gleap agent can do in Metabase
- Run a saved question: Run Card Query executes a saved question, for example an account’s usage by week, with the parameters the agent passes.
- Find the right report: Search Metabase objects finds cards, dashboards and collections by name.
- Understand a question first: Get Card by ID and Get Collection Items show what a question measures before the agent quotes it.
- Get a pivot table: Run Card Pivot Query returns a saved pivot, such as usage by plan and month.
Example: Usage numbers for a renewal call
A customer success manager prepares a renewal call and asks the agent in the customer’s ticket for their usage trend. The agent searches Metabase for the Account usage question, runs it with the customer’s account ID and summarizes the last six months in the ticket. The manager decides how to present the numbers on the call.
How to connect Metabase
- In Gleap, open AI & Automation > Agents and select a custom agent, or create one.
- In its Tools section, click Add tool and search for “Metabase”.
- Pick Metabase under More integrations and click Connect Metabase.
- Enter the Metabase details Gleap asks for, such as your instance URL and an API key, and connect.
- Open the new Metabase tool and switch off every action the agent does not need, such as Delete Card or Delete Dashboard.
- Tell the agent in its instructions when to use Metabase, then test it in chat.
Custom agents are included in the Team, Pro and Enterprise plans. Kai Support can use connected apps as tools on Team, Pro and Enterprise. Kai Code, also on those plans, can use Metabase in coding sessions once you add it under Settings > MCP in Kai Code.
FAQ
Does Gleap integrate with Metabase?
Yes, through Composio. A Gleap custom agent can search Metabase, read cards, dashboards and collections and run saved questions. Results come from Metabase at the moment the agent runs.
What can a Gleap AI agent do in Metabase?
It answers account and usage questions from tickets with your saved Metabase questions, so support quotes the same definitions as your analysts. It can also explain what a question measures.
Which Gleap plan includes the Metabase integration?
Kai Support and Kai Code can use the Metabase toolkit on Team, Pro and Enterprise; custom agents that use it are on Team, Pro and Enterprise. Kai Support answers customers, so switch on only actions that are safe in a customer conversation. The Metabase API key you create decides which collections and data the agent can see. AI usage is billed separately; see pricing.
Which Metabase actions should we switch off?
Anything that deletes or changes content, such as Delete Card and Delete Dashboard, plus admin actions you do not need. The toolkit exposes a large part of the Metabase API, so start with search and saved-question actions.
