I spend my days building software, and lately I’ve had a couple of new coworkers: GitHub Copilot and Claude Code. They’re fast, enthusiastic, and occasionally very confident about code that absolutely makes no sense. I’ll share the wins, the weirdness, and the moments when “AI pair programming” felt more like supervising a very fast intern who’s read the whole internet but hasn’t met my codebases. This is my personal journey with GenAI coding tools: learning how to ask, when to trust, when to double-check, and when to say, “Thanks, but I’ll take it from here.” Expect practical lessons, honest struggles, and hopefully fewer bugs than my AI assistants have suggested.

Using AI to write code is fun; putting AI inside an application for customers is where the fun gets… awesome. I’ll share my adventures building GenAI-powered applications with Microsoft Foundry, Semantic Kernel, and the Microsoft Agent Framework: Where prompts become features, agents get jobs, and suddenly everyone has opinions about orchestration. We’ll explore what it takes to turn AI from a clever demo into something useful for real people, including the choices, surprises, and occasional “maybe the model shouldn’t be in charge of that”-moments. Just practical experience, lessons learned, and a few stories from the frontier of making software talk back.

I'll answer all the questions you post during the show.