Tag
18 posts
August 13, 2026
Mike has been building apps on Fabric as a backend since January, and he's coining it FAB. The case isn't that you already own Fabric — it's that the workspace identity, git integration, and SQL database remove the wiring you'd otherwise do by hand in Azure.
August 11, 2026
Meagan Longoria's five-stage maturity model for agentic Power BI development gives Mike and Tommy a framework to argue with. They mostly agree — except on the crawl stage, where Mike thinks starting with an MCP server beats starting with a chatbot.
August 6, 2026
Desktop Bridge gives agents a local server to drive Power BI Desktop — reload files, check report state, and take screenshots so the agent can verify its own work. Mike and Tommy agree on the prerequisite: don't use it without skills.
July 30, 2026
Microsoft is a Leader in the 2026 Gartner Magic Quadrant for Analytics and BI — first on execution, fourth on vision. Mike and Tommy read the strengths and cautions closely, and land on why "throw AI at your data and ask it questions" is the wrong story.
July 28, 2026
A listener watches junior staff prompt their way to DAX they don't understand, and asks how to train them. Mike and Tommy name the three fundamentals they'd require before handing over agentic tooling — and argue the senior's real job is now building the agent that teaches everyone else.
July 23, 2026
AI-written pipelines don't fail the way human-written ones do — no red squiggles, no thrown exceptions, generous comments, and wrong data. Mike and Tommy work through Nikola Ilic's specify, bound, validate framework and what it means for who you let near an agent.
July 21, 2026
Mike is deleting paid subscriptions and rebuilding them as Fabric apps; Tommy wants to know who maintains all this and whether enterprise security teams will ever allow it. Plus the licensing discovery of the episode — agents talking to semantic models over XMLA with no Power BI Pro license required.
July 14, 2026
Agents will not run your data governance program — most of governance is a committee, not a computer. What they will do is the mundane discoverability work: cataloguing models, surfacing ambiguous definitions, and making lineage human-readable.
July 9, 2026
A mailbag question about request tracking turns into the clearest description yet of what a working harness contains: skills, agents, organized context, and the ability to talk to other harnesses. Plus Mike hits real Rayfin limitations trying to build a simple forms app.
July 2, 2026
A mailbag question about 30-60-90 day AI plans turns into an honest accounting of why organizations are stuck: IT doesn't know what an agent is, nobody has bought into the value yet, and there is essentially no training. Mike and Tommy separate AI strategy from AI theater and each give a real 30-60-90.
June 25, 2026
Part two picks up at the skills and validation layers of Anthropic's analytics stack, and Tommy arrives at a hot take he'll defend: no agentic solution on your Fabric data until governance is in place. Along the way — why perspectives are Power BI's most untapped feature and why Microsoft needs skills for models.
June 23, 2026
Anthropic published how they run self-service analytics internally, and the answer starts with dimensional modeling and a semantic layer — not with the agent. Mike and Tommy read it as validation that everything Power BI professionals already do well is now the thing that makes AI work at all.
June 18, 2026
Agent skills live on your laptop, which is fine until three people on your team each have their own version of the same one. Mike and Tommy work out which skills should be governed, which should stay personal, and why nobody has a good answer yet for measuring whether a skill is any good.
June 16, 2026
If agents write the DAX and the Python, the job shifts from producing code to vetting it — and that makes you a professional reviewer. Mike and Tommy work through what still has to be in your head to review well, and the uncomfortable question of how juniors become seniors when the MCP server is doing the junior's job.
June 11, 2026
Rayfin lets you build full applications on top of Fabric, and Mike calls it the biggest Microsoft release since Power BI Desktop. This episode covers what Rayfin actually is, why the semantic model matters more than ever, and the handoff problem nobody on your team has solved yet.
June 10, 2026
Mike and Tommy recap Microsoft Build 2026 through a Power BI and Fabric lens, focusing on the announcements that matter most for report builders, semantic models, and agent-driven app development. They unpack why Rayfin, Replit integration, OneLake, and new Fabric skills signal a shift from asking AI for answers toward using agents to create real production artifacts.
June 5, 2026
Mike and Tommy explore how agents can simplify CI/CD in Fabric by helping teams work through Git workflows, merge conflicts, validation, and deployment automation. They also compare deployment pipelines with Git-based approaches and explain why small, testable automations are the safest way to start.
April 21, 2026
A hard look at why rigid SaaS stacks are breaking down, how AI agents are becoming the new interface layer, and what that means for teams building on Microsoft Fabric and Power BI.