Now that everything is agentic, what should I be learning?

Filisha Shah explains what to focus on as more application logic shifts into AI agents, highlighting why observability and evaluation become core skills and how Azure AI Foundry brings datasets, evaluations, and monitoring together for keeping agentic systems healthy.

Overview

As traditional application logic moves into AI agents, the video focuses on the practical skills needed to operate these systems in production.

Key points covered:

• Observability for agents: because agent behavior is non-deterministic, keeping systems reliable requires stronger monitoring and diagnostics than typical deterministic app flows.
• Evaluations and evaluators: the video calls out learning how to run evaluations and build/use evaluators to measure agent quality and health.
• Azure AI Foundry as the platform: Azure AI Foundry is presented as a place where datasets, evaluations, and observability are brought together in a single workflow for building and operating agentic applications.