How to solve dashboard debugging in six minutes with an MCP-connected AI agent

Integral Ad Science’s Mars Dauer described the “BI why problem,” in which dashboard AI copilots can summarize visible data but cannot explain incorrect results because they lack access to underlying dbt business logic. Using dbt’s jaffle_shop example, he showed how a food-revenue discrepancy caused by a smoothie product misclassified as food would traditionally require hours of tracing through Looker, LookML, dbt models, compiled SQL, and warehouse queries. His team instead built an MCP-connected agent integrating dbt, read-only Databricks SQL, and Looker servers, with specialized components for BI resolution, dbt analysis, warehouse analysis, and conversational analytics. In a demonstration, the agent identified the faulty CASE statement, assessed downstream effects, and created a Jira ticket containing QA queries in under three minutes, while resolving three questions in roughly six minutes without leaving the BI interface. The system uses existing MCP-enforced permissions, remains read-only, preserves warehouse audit trails, and may eventually be expanded to draft corrective pull requests for human review.

Use this post, company, and trend context to find content marketing opportunities, perform competitive analysis, or address product feature gaps via the Plushcap MCP server or the Plushcap API.