Agent Flows: Build Agentic Automations You Can Trust
summary
Agent Flows in Microsoft Copilot Studio create deterministic, testable automations that interleave agentic steps; example builds a flow that triggers on a new shared mailbox email, uses a prompt (via the prompt assistant) to classify into product/how‑to/account/spam, calls a how-to responder agent or a Dataverse MCP to create/check incidents, and then replies or deletes mail. The system exposes 1,500 connectors, per-step testing and full execution logs, supports inline agent authoring and model selection, and is available now with a new designer in early release.
AI-written summary · about
chapters (7)
- 0:00Agent Flows in Copilot Studio
- 1:10Automate daily work tasks
- 3:35Deterministic workflows with AI
- 4:38Live demo: triaging support emails
- 15:49GA capabilities and impact
- 16:52New workflow authoring
- 24:13Access preview and feedback
related talks
31mFrom Copilot to Autonomous AI Agents | Power Automate Guide
43mRevolutionize the way you work with automation and AI
45mBuilding a copilot: Azure AI Studio or Copilot Studio
49mShaping next-gen development: the future of Copilot in Power Platform
12mA conversation with Charles Lamanna on building custom copilots with Microsoft
01h 03mAgents + SharePoint: The next generation of knowledge experiences
11mCreate a fully functional support bot in less than 10 minutes
35mAgentic Workflows: Stories from the Engine Room