How to solve dashboard debugging in six minutes with an MCP-connected AI agent
Last edited on Oct 26, 2026
Mars Dauer, a senior director of enterprise data and AI at Integral Ad Science, opened with what he calls the "BI why problem": AI copilots bolted onto dashboards are good at summarizing a chart and listing filters, but they have no way to explain why a number is wrong.
Dauer illustrated the gap with dbt's own jaffle_shop demo project. A dashboard shows food revenue running high, and drilling in turns up a product called Tangaroo classified as food when it's actually a smoothie.
Asking the dashboard's native AI why just returns the existing (wrong) tags back as an answer, because the copilot only has access to the BI layer's gold tables, not the business logic sitting in the dbt models underneath. The old fix was a long manual trace: Looker to LookML to dbt to compiled SQL across several models, then validation queries in the warehouse, sometimes taking a whole afternoon.
Dauer's team built one agent connected to three MCP servers: the dbt MCP server for lineage, model metadata, and compiled SQL; a read-only Databricks SQL MCP server governed by Unity Catalog; and a Looker MCP server for BI access.
Behind a router, the agent splits into four specialists, a Looker resolver, a dbt analyst, a Databricks analyst, and Looker's own conversational analytics, and can answer in an "analyst mode" with full technical detail or a "business mode" in plain English depending on who's asking.
In a live demo, the agent walked the smoothie misclassification back to a CASE statement, checked what else would break if the logic changed, and filed a Jira ticket with QA queries already attached, all in under three minutes. Two more investigations followed in the same session, for three answered questions in about six minutes total, without anyone leaving the BI layer.
Governance came along for free: because the agent inherits whatever access each MCP server already enforces, it's read-only end to end and can't surface data a user couldn't already see, and every query it runs lands in the warehouse's audit trail.
Dauer closed by noting his team is already thinking about the next step: having an agent draft the fixing pull request itself, with a human reviewing before it merges.
Watch every keynote and breakout from dbt Summit 2026 on demand. Watch dbt Summit 2026 on demand →
Get started in dbt
Join the analytics engineers building data infrastructure that actually scales.
Install dbt Wizard CLI
Get started with an agent purpose-built for analytics engineering. It knows which tool to call, which context to pull, and checks its own work before surfacing anything to you.
