AI agents have moved beyond being useful tools to becoming real collaboration partners. However, most of us are still learning how to work with agents through a slow process of trial and error -- rediscovering practices that others have already mastered.

In this session, we'll learn best practices for every layer of an agentic system. First, we'll start inside the agent with its model, context, memories, and skills. Then, we'll work our way out to the agent's harness, workspace, tools, and artifacts. Finally, we'll step outside the agent to cover human-agent collaboration, verification, and alignment.

By the end of this presentation, you'll have a practical set of best practices for AI agents -- and a simple mental model you can apply to any agentic system.