Complete AI Training
Sign inGet my AI kit

Your job's AI kit

Get your AI kit

Tell us who you are and what you do. We show you your kit right away and email you the link: skills, prompts, AI agents, MCP servers and courses for your job.

500+ jobs ready, and we make a kit for any other job. No payment needed to look.

Share

AI agent for business intelligence analysts

Dashboard Data Trust Agent

Dashboards that are either correct and current, or clearly marked when they are not

Dashboard Data Trust Agent: what goes in, what the agent does and what you get

What it does

Dashboards depend on data feeds that can break, lag or change shape without warning, and users lose trust when numbers are quietly wrong. Each morning and after each refresh this agent checks that every feed loaded on time, row counts sit in the normal range, and key totals match a control source. It also checks for schema changes, like a renamed column, that would break a chart. When something is wrong, it works out whether it is a late feed, a data issue or a broken field, and reruns the refresh when that is safe. If the rerun does not fix it, it marks the affected tiles as not current and alerts the owner. You approve any banner shown to users. Edge case: a real drop, like a slow holiday sales day, is not flagged after checking against last year.

How it works

Follow the arrows from top to bottom. The orange dashed arrow is the loop: when a check fails, the agent goes back and tries again.

Start and resultWhat it doesA check on its own workWaits for your OKGoes back and retries
Yes, continueYes, continueApprovedNoNo 1 STARTS WHEN Scheduled refresh finishes 2 USES A TOOL Check feed load times, row counts and schema forevery source 3 DOES Compare key totals with the control source 4 DOES Compare any unusual drop with the same period lastyear 5 CHECKS THE RESULT Is every feed current, complete and matching controltotals? If not: identify the late, broken or wrong feed andrerun its refresh. Back to step 2. 6 CHECKS THE RESULT Did the rerun fix the feed? If not: mark the affected tiles as not current and alertthe dashboard owner. Back to step 3. 7 DOES Draft a short health note and a user banner ifneeded 8 YOU APPROVE Owner approves any user-facing banner 9 RESULT Dashboard health log updated
Read the steps as a list
  1. Scheduled refresh finishes
  2. Check feed load times, row counts and schema for every source
  3. Compare key totals with the control source
  4. Compare any unusual drop with the same period last year
  5. Is every feed current, complete and matching control totals?If not: identify the late, broken or wrong feed and rerun its refresh. Back to step 2.
  6. Did the rerun fix the feed?If not: mark the affected tiles as not current and alert the dashboard owner. Back to step 3.
  7. Draft a short health note and a user banner if needed
  8. Owner approves any user-facing bannerThe agent waits here for your OK.
  9. Dashboard health log updated

How it decides

It compares loads and totals with normal ranges and control sources, and marks tiles stale when it cannot confirm they are current.

  • Mark tiles stale when data cannot be confirmed
  • Separate a real low day from a broken feed
  • Rerun a refresh at most twice before escalating

Make it yours

Every agent is a starting point. You choose these settings for your own situation.

  • Control totals and tolerance
  • Normal row count ranges
  • Dashboards in scope
  • Who is alerted

What keeps you in control

It always asks you first

  • Showing a banner or notice to dashboard users

Hard limits

  • Never changes dashboard numbers manually
  • No user banner without approval

It stops when

  • Done: dashboards verified or marked stale
  • Stop: pipeline logs unavailable

Set it up

We guide you through the set-up, step by step

Members get the full set-up guide for this agent. No technical skills needed: you copy, paste and upload.

10 minto set it up in your AI
5 AIsChatGPT, Claude, Copilot, Gemini, Grok
  • One set of instructions to paste into your AI, with the clicks for ChatGPT, Claude, Microsoft 365 Copilot, Gemini and Grok
  • The agent then walks you through connecting your own data, one source at a time
  • A downloadable copy with the flow chart, the rules and the full guide
Get access to this agent

An example run

What happensOn Monday at 07:00, the Brookside Foods sales dashboard showed 30% fewer orders. The agent found the orders feed had loaded only half its rows, so the completeness check failed. A rerun loaded all rows and totals matched the control source. A low count on the web dashboard matched last year's holiday dip, so it was left as normal. The BI owner approved no banner.

More agents for business intelligence analysts