The curriculum
Everything, in the order it makes sense to learn it.
14 chapters and 101 lessons covering the whole product management craft plus the AI layer that now sits under it. Each lesson is one short video, one written lesson, and something you do. Every chapter ends in a graded build set inside a real product, a different one each chapter, so you finish with a portfolio across the landscape rather than sixteen pieces of work in the same app.
This curriculum is not fixed, lessons can be rewritten, added or retired as the job and the tools change.
01 Foundations
Foundations: the model and the decisions it forces on you
Technical fluency and ML tradeoffs is the single strongest theme (90 percent of JDs, 77 companies), so the course opens here.
- 01.01 The AI PM job now11 min · Free
- 01.02 Make the API call yourself14 min · Free
- 01.03 One token at a time12 min · Free
- 01.04 Context windows and attention14 min · Free
- 01.05 Pretraining vs post-training16 min · Free
- 01.06 The AI-or-not decision14 min · Free
- 01.07 Read a benchmark honestly17 min
- 01.08 Attribute every failure to a layer12 min
- 01.09 INTEGRATION CASE120 min
02 Data fluency
Data fluency: SQL, logs, and reading the truth yourself
The report names SQL as the single most-named tool (27), ahead of APIs (24) and MCP (23), and flags that no course teaches it: this is called out twice, in the missing-material list and as one of the sharpest JD-versus-course disagreements
- 02.01 SQL for PMs18 min
- 02.02 Read the logs16 min
- 02.03 Cohorts20 min
- 02.04 From raw data to a golden sample16 min
- 02.05 INTEGRATION CASE22 min
03 PM as builder
PM as builder: prototype and inspect the agent yourself
PM-as-builder with agentic coding tools is a named gap and a named frontier.
- 03.01 Build and iterate in Claude Code16 min
- 03.02 Read the code you did not write18 min
- 03.03 MCP first contact15 min
- 03.04 Prototype by what it must prove14 min
- 03.05 One output is an example; ten inputs by five r16 min
- 03.06 INTEGRATION CASE24 min
04 Discovery and strategy for AI products
Discovery and strategy for AI products
AI PM craft is the second strongest theme (88 percent, 73 companies) and building under ambiguity is 68 percent, both core.
- 04.01 Product sense under ambiguity14 min
- 04.02 Discovery for AI12 min
- 04.03 The problem-first test16 min
- 04.04 Qualify the opportunity14 min
- 04.05 The capability horizon18 min
- 04.06 Goals15 min
- 04.07 INTEGRATION CASE120 min
05 The AI PRD
The AI PRD, the spec, and the road to launch
Still inside the 88 percent craft theme, this chapter owns the artifact the report calls the differentiator from classic PM: the AI PRD must name risks, guardrails, and success metrics up front because the system is non-deterministic (Produ
- 05.01 The AI PRD18 min
- 05.02 Roadmap around an evaluable slice15 min
- 05.03 Write the spec after the demo22 min
- 05.04 From agreed spec to shipped14 min
- 05.05 Launch readiness and the narrative16 min
06 Agents and agentic architecture
Agents and agentic architecture
Agents are 73 percent (68 companies) and enterprise AI agents are the single top product surface (11), so this is a central, mid-course chapter.
- 06.01 Workflow or agent14 min
- 06.02 The anatomy of an agent18 min
- 06.03 The agent spec you own16 min
- 06.04 Write the tool contract22 min
- 06.05 Inside the harness20 min
- 06.06 Context as a budget18 min
- 06.07 When multi-agent is worth it22 min
- 06.08 Reachable from someone else's agent18 min
- 06.09 Structured outputs are a decoding constraint16 min
- 06.10 INTEGRATION CASE20 min
07 AI UX and human oversight
AI UX and human oversight: design for a system that is wrong sometimes
AI UX and human oversight is 78 percent (71 companies), core, and central to the mid-course spine.
- 07.01 The four AI design patterns for a surface that14 min
- 07.02 Levels of autonomy18 min
- 07.03 The four surfaces16 min
- 07.04 Stream the work16 min
- 07.05 The approval gate14 min
- 07.06 Claim-level citations15 min
- 07.07 The empty box14 min
08 Evals
Evals: define good and make the number defensible
Evals are 53 percent (62 companies), real but middling against the top themes, and the most over-taught topic in the field.
- 08.01 Read one hundred real traces before you buy an14 min
- 08.02 Name the failure from one taxonomy16 min
- 08.03 Turn one complaint into thirty golden examples12 min
- 08.04 Pay for the cheapest evaluator that can see th14 min
- 08.05 Write the criterion15 min
- 08.06 Validate the judge16 min
- 08.07 Eval-driven development14 min
- 08.08 Evaluate an agent18 min
- 08.09 INTEGRATION CASE24 min
09 Prove it paid off
Prove it paid off: outcomes, economics, and pricing
Outcomes and metrics for AI is 62 percent and, tellingly, outranks evals (53 percent), yet the report says course metric content is dominated by eval mechanics: this is a deepen theme and a named gap.
- 09.01 Business outcomes14 min
- 09.02 Build the metric tree15 min
- 09.03 Pull the number yourself13 min
- 09.04 Diagnose a drop when the treatment is nondeter16 min
- 09.05 Cost per successful task18 min
- 09.06 Price it17 min
- 09.07 The business case16 min
- 09.08 INTEGRATION CASE18 min
10 Ship it into somebody else's company
Ship it into somebody else's company: enterprise brownfield deployment
Enterprise and deployment is 79 percent (70 companies), enterprise AI agents is the top product surface, and the report calls this the largest single mismatch between JDs and courses: the field teaches greenfield capstones while the hiring
- 10.01 Brownfield first14 min
- 10.02 Channels and connectors12 min
- 10.03 Auth and access13 min
- 10.04 The permission-aware retrieval layer14 min
- 10.05 From pilot to production12 min
- 10.06 Procurement12 min
- 10.07 After the signature16 min
- 10.08 Hosted22 min
- 10.09 Who owns the output18 min
- 10.10 INTEGRATION CASE18 min
11 Beyond text
Beyond text: multimodal products
Multimodal is 34 percent (47 companies), which the report notes is broader than several topics that get far more course time, and it is a named gap: nearly absent across the field, with only Product Faculty's Sense layer touching it and no
- 11.01 Documents are not text18 min
- 11.02 Voice20 min
- 11.03 Image and video generation18 min
- 11.04 Ship a language18 min
- 11.05 Build a multimodal golden set16 min
- 11.06 INTEGRATION CASE24 min
12 Trust
Trust, safety, and agent security
Safety, trust and governance is 32 percent, concentrated in only 36 companies, and the report's verdict is right-size: keep guardrails and responsible-AI as a working PM skill (define guardrails in the PRD, require human approval for irreve
- 12.01 Guardrails in the PRD14 min
- 12.02 What an agent may do without asking16 min
- 12.03 Agent security18 min
- 12.04 Guardrails in the request path16 min
- 12.05 Red-team the obvious modes20 min
- 12.06 Right-sized governance14 min
- 12.07 INTEGRATION CASE28 min
13 Lead the room
Lead the room: staff moves, forward-deployed PM, and the portfolio
The seniority mix is senior-heavy (97 senior, 92 mid, 55 staff), so the report's graduate is an experienced PM adding AI depth, and the staff-level moves are part of the target profile.
- 13.01 Decision rights14 min
- 13.02 Work with researchers18 min
- 13.03 Land one operating standard across research15 min
- 13.04 The data flywheel and self-improving agents16 min
- 13.05 Retire your own feature15 min
- 13.06 Capstone24 min
14 Get the job
Get the job: the AI PM interview loop
The report frames the entire course around a demonstrable, hands-on AI PM entering a senior-heavy market (97 senior, 92 mid, 55 staff), where the interview scores evidence of shipping over borrowed vocabulary and the probing question expose
- 14.01 What each round is really scoring14 min
- 14.02 Forty minutes16 min
- 14.03 The take-home18 min
- 14.04 The values round14 min
- 14.05 Read the ladder14 min
- 14.06 Which round lets you use AI10 min