Multimodal in Minutes: Prototyping Agents with Microsoft Foundry + AI Toolkit

Build intelligent, multimodal agents end‑to‑end—faster than ever. In this hands‑on session, Microsoft Solution Engineer Vanessa walks through how to prototype, evaluate, and deploy AI agents using Microsoft Foundry, AI Toolkit for VS Code, MCP servers, and Azure. Learn how to go from idea → prototype → production in under an hour, using real tools, real models, and real workflows designed for developers. 📌 Chapters 00:00 – Opening music 01:22 – Welcome, housekeeping, code of conduct 02:09 – Meet your speaker: Vanessa (Microsoft) 03:12 – How intelligent app development is evolving 04:10 – Generative AI development lifecycle 06:00 – What is the AI Toolkit for VS Code? 08:42 – Use case overview: Building “Kora” for home design 10:00 – Choosing the right model (GPT 5 mini vs 4.1 mini vs 40 mini) 14:53 – Multimodal comparison in Playground 17:10 – Designing effective system prompts 21:00 – Testing and refining agent behavior 27:01 – Adding tools with MCP server integration 33:06 – Agent evaluation (manual + automated) 40:06 – Exporting agent code to run outside VS Code 45:52 – Deploying agent to Azure using Foundry 49:22 – Monitoring agents in Foundry Control Plane 53:05 – Recap, resources, and next steps 56:01 – Q&A and closing remarks ✨ What You’ll Learn How intelligent apps differ from traditional deterministic applications How to select the right language model for your agent Best practices for writing effective system prompts How to integrate dynamic tools with MCP servers How to evaluate agent results with AI Toolkit How to export agent code and deploy it into Azure How to monitor and operate agents using Microsoft Foundry 🎓 Join the AI Genius Series! Scan the feedback QR code from the session Claim your AI Genius Certified badge Explore upcoming sessions in the AI Genius learning path [eventID:26579]