Every week there's a new AI tool promising to change how you build Power BI. Most won't survive contact with real work. I've run them against real models, real deadlines, and real client deliverables, and this session is the verdict.

No survey of the whole field. No hype. Just the tools I actually reach for, and exactly why they earn their place.

We'll climb three rungs, each one a live demo using Claude Code connected directly to a real Power BI model:

Chat-level AI, the ChatGPT and Claude tools you already use, and where they quietly fail you.
Context-aware AI that reads your actual semantic model instead of guessing at it.
Agentic AI that doesn't just answer questions but acts: writing measures, documenting models, and committing changes for you.
You'll leave knowing which rung you're on today, what the next one looks like, and the one concrete step to climb it.

Three takeaways:

The difference between AI that guesses at your model and AI that reads it, and why that's the whole game.
A live look at an agent writing, documenting, and version-controlling a semantic model.
A curated shortlist of the tools worth your time, from someone who's discarded the rest.