Most courses about AI coding tools have the same problem. You watch a video of someone using the tool, then switch windows and try to remember what they did. Carl Vellotti's Claude Code for Product Managers sidesteps that entirely. You install the Claude desktop app, paste a single prompt into the Code tab, and the course installs itself into your workspace.
From then on Claude Code is both the teacher and the thing being taught. Lessons ask you to do real PM work with real files: organise a messy folder, spin up sub-agents, set project memory, write a PRD, analyse a dataset and work through a product strategy exercise. It is the right way to learn a tool like this, and I wish more vendors had thought of it first. Module 0 handles installation, including a separate path for people who prefer the terminal.
Module 1 covers the fundamentals, and Module 2 moves into PM deliverables. The sidebar has since grown to include an image generation module, a five-part vibe coding module and a guide to Cowork, so the course is clearly being maintained. Vellotti puts the time at 10 to 12 hours, which matches what you would expect, and assumes you know what a PRD is but nothing about code. The course itself costs nothing, and Vellotti's only commercial interest is his newsletter and community, The Full Stack PM.
The catch is that Claude Code needs a paid Claude plan, so budget about $20 for the month. That is still far cheaper than any cohort on this topic. Two caveats. First, this is an independent project, and the site says clearly that it is not affiliated with Anthropic.
Claude Code changes quickly and a one-person course will sometimes describe a menu or command that has moved. Second, there is no certificate and nobody to ask when you get stuck, though the tool itself is a fairly patient tutor. The licence is non-commercial with no derivatives, which is fair for a free course but means a product org cannot fork it into its own onboarding. My view is simple: this should be the first stop for any PM curious about agentic tools, and it sets a useful bar for paid courses.
If a $3,000 cohort does not teach you meaningfully more than this free one, it is not worth the money.