This session reviews a Personal Health Agent hosted in AKS that supports both Microsoft AI Foundry hosted LLMs and AKS KAITO hosted LLMs.
A demo will cover scenarios of storing health profiles and photos of meals and infer macronutrients and store in an Azure SQL DB for long-term agent memory. Then ask questions to the agent about past health history, historical calorie deficits, health goal tracking for weight loss, and much more.
It covers the application architecture, Microsoft Agent Framework orchestration, Streamlit front end, durable memory design, and model selection across cloud and AKS in-cluster backends with KAITO LLM hosted models such as phi 3.5 vision.