You've been using agents for a while now, Claude Code, Codex, the usual suspects. But most of them still live in the terminal and their main goal is writing code.
Claude Cowork is different. It's Anthropic's native desktop agent:
A proper app sitting on your machine with full access to your files, your apps and the tools you've connected to it.
Same model and tool-use primitives you know from the API, but with a real UI and OS-level integration.
Think of it as Claude Code's sibling for everything that isn't code:
Research, docs, file workflows, browser automation, full desktop control.
All in one place, all reachable through the UI.
The course takes you all the way from "what even is Cowork" to "I have a real working Cowork setup tailored to my own work".
Among other things we'll dive into:
The agent mental model and how Cowork differs from regular chat
The interface, permission modes, and file sandbox
Outcome-based prompting that holds up across multi-step tasks
Connectors and MCP to wire Cowork into your real workflow
Skills and plugins for your repeatable processes
Projects with persistent memory and live artifacts
Scheduled tasks, Claude in Chrome, and Dispatch
Computer Use, including the security side most courses skip
