Safer Agentic Workflows | THR1009
Jonah Andersson and Danielle Cass present an intermediate-level look at safer agentic workflows, focusing on guardrails, evaluation, and human involvement.
Overview
The session examines three approaches to helping AI agents stay useful, reliable, and easier to trust:
- Guardrails for agent behavior.
- Evaluation of agent workflows.
- Human-in-the-loop patterns that keep people involved in those workflows.
Session details
- Event: Microsoft Ignite 2026
- Session: THR1009 — Agent Platform
- Format: Theater
- Level: 200 — Intermediate
- Speakers: Jonah Andersson and Danielle Cass