Building an AI solution in 2026 means facing a galaxy of models, SDKs, frameworks, and patterns that did not exist a year ago, and a fair amount of it will have changed again by the time you get home. Don't panic. In this full-day workshop, Brian and Matt hand you the field guide: a practical tour of the patterns that actually matter for shipping an AI solution on the Microsoft stack, grounded in live demos you can take home as working reference code.

Rather than building one monolithic app, Brian and Matt walk the whole map, stopping at each pattern a real AI solution needs and showing it in code. You will move through choosing and accessing models with GitHub Models and Microsoft Foundry, grounding answers in your own data with retrieval patterns, adding agents and tool use with the Microsoft Agent Framework, and connecting external capabilities through MCP. From there the day turns to the parts that separate a demo from a solution: evaluating and observing what your AI actually does, keeping it secure and responsible by design rather than as an afterthought, and getting it to production with GitHub and Azure. Throughout, Brian and Matt stay honest about cost, tradeoffs, and where each pattern fits, so you leave knowing not just how to build, but what to build and when.

You will learn:

• How to choose and connect to models, ground them in your own data, and add agents and tool use on the Microsoft stack, using the current SDKs and frameworks rather than deprecated ones
• The patterns that turn a working demo into a real solution: evaluation, observability, security by design, and deployment with GitHub and Azure
• A clear-eyed map of the tradeoffs, including cost and hosting choices, so you can decide what to build and what to skip for your own solution