In this session you will learn how to build a fully functional AI-driven coaching application using Azure AI Foundry’s latest capabilities. We will present an end-to-end architecture based on Microsoft Agent Framework (public preview) — a unified, enterprise-grade SDK and runtime that consolidates strengths of previous frameworks (e.g. AutoGen and Semantic Kernel) and enables both multi-agent orchestration and structured, long-running workflows.

The demo will showcase:

- Voice-enabled, expressive agents built with Azure AI Speech, including photorealistic avatars

- Integration with Azure OpenAI Service to power natural-language reasoning and generation under the hood.

- Advanced monitoring, evaluation and observability capabilities powered by Built-in “Tracing, Evaluation and Monitoring” (AgentOps) features — enabling transparent evaluation of agent behavior (intent resolution, tool-call accuracy, task adherence, etc.) and full traceability of agent workflows in production.

This provides a practical blueprint for delivering a secure, scalable, human-like coaching experience — from prototype to production — with enterprise-grade quality, governance, and observability baked in.

To encourage live participation, small tokens of appreciation will be offered for the best questions during Q&A.