Supercharge Microsoft Power BI Development: How AI and TMDL Transform Your Workflow

Using AI to Understand Your Microsoft Power BI Model

So, by using a TMDL viewer, you can expose your model’s structure – including tables, columns, relationships, and measures – in a format that AI systems can easily interpret. This opens the door to using AI not just for ideation, but for actual development tasks.

It becomes a simple task to copy and paste the TMDL code into your AI system of choice. One caveat to be aware of is that, depending on what object(s) are being described, the TMDL view can become several thousand lines long. All AI systems will have limits in terms of how much data they can work with. These limits are much lower with free AI systems. However, my experience is that copying and pasting the TMDL view into notepad and uploading the resultant file provides a much wider scope for larger amounts of text and often remain within the limits of free AI services.

Claude Sonnet: A Strong Performer for Microsoft Power BI Tasks

At the time of writing, Claude Sonnet is showing excellent performance when working with TMDL inputs. However, I have found that Copilot with GPT-5 can give helpful results as well. With this being such a dynamic area, this situation will change day by day, so test out the systems you have access to, to work out which one works best for you.

Once you have your TMDL view data, you can use this in combination with a well-formed query to generate new DAX measures, suggest improvements, or even refactor existing logic. The results are often well-structured, readable, and ready to paste directly into your TMDL sheet.

“Please review the following TMDL file, which represents the semantic model of my Power BI report. I’d like you to analyse it and provide suggestions for improving the data model. Focus on areas such as:

• Table and column organisation
• Relationships and cardinality
• Naming conventions and consistency
• Use of calculated columns vs measures
• Performance optimisation opportunities
• Star schema design and granularity
• Any redundant or unused elements

Please explain your recommendations clearly and reference specific parts of the model where applicable.”

Generating DAX Measures with AI

One of the most practical use cases is generating DAX measures. Writing effective DAX measures can be troublesome, and even experienced Microsoft Power BI developers can sometimes struggle with this aspect. With the right prompt and context from your TMDL viewer, AI can:

• Create new measures based on business logic or KPIs
• Format them correctly for TMDL inclusion
• Include display folder assignments and formatting strings

This means you can go from idea to implementation in seconds – without switching between tools or manually formatting your code.

Creating a measure where you know what you need is one approach. However, what also works surprisingly well is to provide the AI system with the semantic model and ask it to generate measures that may be useful. This will not only create many measures you would have had to build one by one yourself, but also often creates measures you hadn’t considered, and which might prove extremely helpful.

“Using the TMDL definition, create appropriate measures in TMDL format (Indentation is important) that may be useful with this report, using the example below. Include a definition for each measure using /// to indicate a comment. When the measure is a number, format it with a thousand separator and no decimals. When it’s a percentage, format it as a percentage with 1 decimal place.

createOrReplace

    ref table Orders

        measure ‘Sales measure’ = sumx(Orders, Orders[Quantity]*Orders[UnitPrice])

            lineageTag: 948d6cfe-89d8-44fa-87f1-79d1600b20b8

            annotation PBI_FormatHint = {“isGeneralNumber”:true}”

Conclusion

The TMDL view is ideal to provide the context that AI systems need. The combination provides scope for dramatic improvements in your reports and reductions in development timelines. Start experimenting with TMDL and AI today – your next Microsoft Power BI report could be your fastest yet!