As of 13 August 2026, AI can only partly build a Power BI dashboard from your data.

What to actually do

1. Hand it to a person

   The route this page recommends

   A person who owns the outcome does this end to end, worth it when the failure is dear.

2. Use a tool built for this

   Second choice

3. Do it yourself

   The distant third

   A chat interface, power-user skill, and roughly 1 hour until you can act on the result.

   How to actually do it

   1. Open the source files and remove unnecessary personal or confidential fields, then create a small representative sample with the same column names and data types.
   2. Write down the business purpose, intended users, metric definitions, required filters, reporting period, refresh frequency and any row-level access rules.
   3. Paste the prompt into an AI chatbot, attach the representative data, and replace each bracketed slot with your reporting requirements.
   4. Check the AI's table, column, data type and relationship assessment against the source files before using its Power Query or DAX output.
   5. Open Power BI Desktop, import the data, apply the proposed Power Query steps and create the relationships and measures one at a time.
   6. Build the proposed report pages and compare every key total, date filter and category breakdown with independently calculated totals from the source data.
   7. Ask a colleague who owns the reporting process to test the dashboard with realistic questions and confirm the figures, permissions, refresh behaviour and interpretation before publishing.

   Prompt

   I need to build a Power BI dashboard from the attached data.
   
   Business purpose: [describe the decisions this dashboard should support]
   Audience: [name the teams or roles who will use it]
   Required measures: [list each metric and its exact business definition]
   Required filters: [list the filters users need]
   Reporting period: [state the date range and time grain]
   Refresh requirement: [state how often the data should refresh]
   Security requirements: [state who may see which rows or fields]
   
   First, inspect the data and report the tables, columns, data types, missing values, duplicate keys, likely relationships and any ambiguity. Do not invent business definitions, values or relationships. Ask targeted questions where the information is missing.
   
   Then produce a Power BI build plan containing:
   1. A proposed star schema and relationship settings, with the reason for each relationship.
   2. Power Query transformation steps, written so I can apply them in Power BI.
   3. DAX measures for the requested metrics, with plain-English explanations and assumptions.
   4. A page-by-page report layout with recommended visuals, fields, filters and titles.
   5. A validation plan with test cases that compare dashboard results with independently calculated source totals.
   6. A privacy and access checklist, including row-level security where relevant.
   
   Use only the supplied data and definitions. Mark every assumption clearly. Do not claim that the dashboard has been built or tested. End with the smallest set of actions I must complete in Power BI to implement and publish it.

   Open it prefilled in ChatGPT or Claude, or copy it into Gemini, which takes no prefill link.