In this episode of Dynamics Corner, Kris and Brad welcome back Torben Leth, Microsoft MVP and creator of CentralGauge, for a follow-up conversation about how much has changed in just a few months. A lot has changed. Torben's benchmarks that used to separate the good models from the great ones? The models blew right past them. They're so good now that he has to completely redo his test sets just to find meaningful differences. The real gem in this conversation is Torben's Eureka moment for any AL developer getting started with AI. It's not fancy tooling. It's not the perfect model. It's the first time you get your setup to automatically compile, deploy, and run your tests. That's when it clicks. That's when you see how fast the iteration loop can be and realize you're not going back. From there, Torben walks through his full multi-model workflow: Opus handles the planning, Fable steps in as the "talking to an adult" for spec reviews, and Sonnet does the actual coding because the spec is solid enough that it no longer needs a heavyweight model. He's even pulling in GPT and Gemini through an MCP called PAL to get a third and fourth opinion before anything gets built. Then there's the story about his Linux laptop. He needed a fingerprint sensor driver that didn't exist. So, he pointed AI at Wireshark, had it capture the USB traffic from Windows, and let it work overnight. By morning, he had a working Linux driver. He didn't write a single line of it. The conversation wraps with where costs are headed. Torben's prediction? Premium models will handle the thinking, and local open-source models will handle the coding. Whether you've been deep in AI development for months or you're still figuring out where to start, Torben's Eureka moment might be the push you need.

Send us Fan Mail

Support the show

#MSDyn365BC #BusinessCentral #BC #DynamicsCorner

Follow Kris and Brad for more content:
https://matalino.io/bio
https://bprendergast.bio.link/