Everything an AI PM knows at Anthropic

From the market. Still moving.

Built from 604 job postings across 95 companies and 137 careers boards, this course follows real hiring data, not one instructor's opinion. New AI and agent concepts keep appearing, and the course folds them back in, so unlike a normal course, it does not go stale.

The course, and everything around it.

One account: the AI PM course, unlimited practice built from real job descriptions, and the libraries and tools product managers actually use.

The AI PM knowledge graph

Explore 212 concepts and 1,113 connections as a 3D galaxy or flat 2D view. Sign in to light up finished concepts. Click any star to see what it is, why it matters, the lesson, and its connections.

604 job descriptions, read in full, at 95 companies.

We pulled every Product Manager listing we could reach across 137 careers boards, 570,000 words of them, then built the curriculum backwards from what they actually ask for. Every chapter traces to a line somebody is hiring against.

This course is built differently for three reasons.

The market sets it. Sources ground it. Real products turn ideas into judgment.

The market sets the syllabus.
We started with 570,000 words from 604 roles across 95 companies and 137 careers boards. That hiring evidence decides what gets taught, not one person's taste.
The lessons go back to the source.
You start with the original talk, interview, or demo, often at the exact timestamp. The written lesson explains it, cites the source, and is not a thinner repackaged lecture.
You practice inside real products.
Thirteen of fourteen chapters end with case studies in NotebookLM, Notion, Perplexity, Cursor, Figma, or ElevenLabs. The hiring chapter ends differently: you submit a PDF, and the model scores it against a rubric, quotes lines, names gaps, and shows a better rewrite.

Here is what a chapter looks like.

Start with the source, move to the explanation, then use the idea in a product you can inspect.

  1. Watch. You start with the original source, often at the exact minute in a talk, interview, or demo from its builders.
  2. Read. The lesson note explains the idea in plain language, why it matters to a PM, and cites every source behind the claim.
  3. Apply. You answer a PDF case study in a real product, and the model scores each criterion, cites draft evidence, and shows a stronger rewrite.

The course moves with the field.

Weekly, we reread podcasts, live job postings, and model provider docs. New concepts can enter the AI PM knowledge graph, lessons can be rewritten or retired, and every change is public and dated.

Podcasts and leader talks
PM and AI leader talks refresh lessons and add concepts when new ways of building appear.
Live job postings
Live roles reveal new skills companies want, so we shift the curriculum toward the market you will face.
Model provider docs
New provider capabilities or limits update examples, concepts, and lessons before guidance goes stale.

You buy at the early price, get the complete course today, watch it improve in public, and verify it in the public changelog.

What people ask before starting.

Do I get the whole course right away?
Yes. 14 chapters, 101 lessons, and 31 hours are live now. Thirteen of fourteen chapters end in a graded case study. The hiring chapter is different. What changes after purchase is the material, not whether it exists.
Why take this instead of Maven, Coursera, or YouTube?
A cohort usually teaches one instructor's style, a platform course stays general, and a playlist gives you no path. This course starts from 604 job postings across 95 companies, then turns that demand into a structured path with source material and practice.
Do I need a technical background to follow this?
No AI background is needed. It starts with core product work, then layers AI in the same order, explaining context windows, tokens, cost per million tokens, latency, evals, agents, and retrieval. You do not need prior model provider vocabulary.

The curriculum

Explore