This course contains the use of artificial intelligence.

Generating a PRD with AI takes ninety seconds. Knowing whether it is true takes a method.

That gap is where this course lives. Five other Claude courses for product managers will show you how to produce a spec and a themed summary of your feedback. Not one of them teaches you how to check either. So the PRD ships with an invented approval rule buried in section four, and the synthesis reports that 42% of customers want better reporting when what it actually measured was 42% of the people who bothered to write in.

This course is built on the second half of the job.

What makes this different

Grounding is taught as a habit, not a disclaimer — four moves you make before the question, every time, in Section 2. Feedback synthesis ships with a validation protocol: re-run stability, spot-checks against raw quotes, and the difference between frequency in your corpus and prevalence in your customer base. You will watch Claude invent a policy rule on screen, in a real session, and learn why it read as completely plausible.

There is no product-management fundamentals section

The loudest complaint in the reviews of every competing course is padding — being told what a PM is by someone you came to for something else. PM method appears here only where it is the thing being checked against AI output. Lecture one opens inside a live product problem.

You finish with artefacts, not notes

A grounding and verification checklist. A PRD skill file that triggers itself when the task arises. A feedback synthesis validation protocol. A prioritisation rubric. A one-fact-set sheet that keeps a number identical across five audiences. A PM skill starter kit. A data boundary card. An Article 50 routing card. Everything is built against one company — Fathom, a Berlin spend-management SaaS with 22,000 customers and a flat activation curve — so each artefact connects to the last.

How it is taught

Eight sections, thirty-four lectures, around three hours of video. Seven section assignments plus a capstone. Eight quizzes and a 50-question final practice exam. Eight downloadable templates. Every Claude session you see on screen is a real captured session, not a recreation — a course about Claude should not show you a Claude that isn't one.

One honest note

Model names, tiers and prices move faster than any course can. This one teaches model selection as reasoning — fast tier for extraction, frontier tier for judgement, extended thinking for hard trade-offs — so what you learn survives the next release. Dated regulatory content is confined to a single lecture for the same reason.