Model Studio

AI DAX generator that knows your actual Power BI data model.

Upload your TMDL semantic model and describe the measure you need in plain English. Model Studio generates production-ready DAX using your real table names, column names, and relationships, not generic examples that need rewriting.

Try Model Studio free

Free tier — 5 AI Actions per month, no card required. Works with any Power BI license.

Generic DAX generators get the table and column names wrong. Then you're debugging instead of building.

I've tried using ChatGPT for DAX. It gives me a formula that looks right, but the table is called 'SalesData' in the DAX when ours is 'fact_sales'. I spend more time fixing the references than I saved on writing the formula.

The fundamental problem with AI DAX generation is context. A language model without access to your semantic model has no way to know that your date table is called dim_date and your sales fact is fact_transactions with a column named net_revenue_usd. It defaults to the most common naming patterns in its training data, which are never exactly right.

Model Studio parses your TMDL schema file (the same format Power BI uses for semantic model version control) and builds the AI prompt context from your real model. Every table, column, relationship, and existing measure is available as context when generating new DAX. The result is DAX that references your actual model structure and doesn't need renaming before it works.

Schema-grounded DAX generation with conversational refinement.

Model Studio gives the AI the context it needs to generate accurate DAX, and keeps that context through follow-up refinements.

TMDL schema parsing with model view

Paste or upload your TMDL schema and Model Studio builds an interactive model view showing all tables, columns, data types, and relationships. Every element becomes context for the AI.

AI-generated DAX using real schema names

Describe the measure in plain English ('calculate running total of net revenue by month excluding returns') and get DAX that references your actual table and column names, not generic placeholders.

Conversational refinement (6 turns)

The AI retains context across up to 6 follow-up turns. Ask it to 'add a filter for region' or 'use a rolling 90-day window instead' and the refinement builds on the previous output.

Monaco editor with DAX syntax highlighting

Generated DAX opens in a Monaco code editor (the same editor used in VS Code) with full DAX syntax highlighting, bracket matching, and find/replace. Edit directly before saving.

Personal measure library

Save finalized measures to a personal library grouped by semantic model. Copy a measure expression with one click, or download your full library as a text file for offline reference.

Regenerate with refinements

Not quite right? Provide additional constraints and regenerate. The system includes the original schema context and your previous instructions so the new attempt learns from what didn't work.

Schema context. No data. No credentials.

TMDL contains your model's structure (table names, column names, relationships, and existing DAX) but no actual data rows. Connection strings and credentials in partition blocks are automatically stripped before AI processing and never reach any AI provider. Processed transiently, never stored, never used for training.

What's included in the free tier

TMDL schema upload and interactive model view
5 AI DAX generation actions per month (resets monthly)
Up to 6 conversational refinement turns per session
Monaco editor with DAX syntax highlighting
Personal measure library: save, copy, and download

From TMDL schema to production DAX in minutes.

Upload or paste your TMDL schema
If your Power BI project uses PBIR format, you already have a TMDL file in your repository. Paste the content or upload the file. Model Studio parses tables, columns, relationships, measures, and hierarchies from the schema automatically.
Describe the measure you need
Type a plain English description of what the measure should calculate. Be as specific as you like. “Calculate the 12-month rolling average of order count excluding cancelled orders, filtered to the selected region” and the AI will use your actual schema to produce accurate DAX.
Refine, edit, and save
Review the generated DAX in the Monaco editor. Ask follow-up questions or provide corrections in the chat. When satisfied, edit directly in the editor if needed, then save to your measure library or copy to paste directly into Power BI Desktop.

Stop debugging DAX that references the wrong tables.

Generate DAX measures grounded in your real semantic model. Works with any Power BI license. No Premium or Fabric capacity required. Free tier includes 5 AI Actions per month, no card required.