Claude Cowork: Build Agent Workflows That Pass Audits

Hosted by Mike Wheeler

116 students

In this video

What you'll learn

Distinguish what Cowork can do, and where it hits limits

Evaluate the trade-offs: clicks over code, local file access, but no cross-session memory. Know before you build.

Design checkpoints that trigger before context runs out

Plan triggers based on time, milestones, and remaining capacity. Capture critical context before limits hit.

Convert your best prompts into reusable skills files

Transform what works into portable .md files you can reuse across Claude Code, Cowork, and other AI tools.

Why this topic matters

Most people struggle to build AI agents that actually work. Those who push forward hit a wall where context quality drops off a cliff. Agents forget prior decisions and handoffs break with no audit trail. But now, you can build agents without IDEs or code thanks to Claude Cowork. I'll show you how to apply agent design and context state management to build agent workflows.

You'll learn from

Mike Wheeler

AI educator and O’Reilly author focused on accountable agent workflow design

I help organizations adopt AI responsibly by designing workflows that persist and scale. After teaching 500,000+ professionals through O'Reilly, Salesforce, edX, and LinkedIn, I've seen countless AI initiatives fail because no one designed for context limits.

Now I teach the two disciplines every AI practitioner needs: Agent Workflow Design (what agents should do) and Context Engineering (what must persist for it to work). My frameworks help leaders build judgment that holds up as tools change.

as seen on