Module 12 — Capstone Project

You made it. The capstone is your opportunity to demonstrate that you can operate as an AI-native PM end-to-end — from discovery to delivery to governance. There is no single right answer. You are assessed on the quality of your thinking, the structure of your system, and the rigour of your process.

Capstone Objectives

Synthesise skills from all 12 modules into a coherent AI-native PM workflow
Demonstrate ability to use AI as an execution partner, not just a query tool
Produce a complete PM OS starter kit for a real or hypothetical product
Show evidence of the QA loop — human review and critical evaluation of AI outputs
Present a governance-ready AI feature design

Capstone Brief

Choose one of the following scenarios. All scenarios use the FlowScale context unless you substitute your own real product.

Scenario A — New Feature Launch: FlowScale is launching a Smart Dunning feature: an AI-powered failed payment recovery engine that personalises retry timing, messaging, and escalation based on each subscriber's payment behaviour. You are the PM.

Scenario B — Own Product: Use a product you currently work on or have worked on recently. Apply the full AI-native PM workflow to one feature or initiative of your choice.

Both scenarios require the same 6 deliverables.

The 6 Capstone Deliverables

Deliverable 1 — Product Context File

A context.md file covering: product vision, primary users, top 3 problems being solved, current success metrics, and key constraints. Under 400 words. Written with AI assistance but edited and verified by you.

Deliverable 2 — Discovery Synthesis

A user research synthesis document based on at least 3 simulated (or real) user interviews. Must include: top pain points ranked by frequency and severity, one opportunity statement in the format "How Might We...", and at least one surprising or counter-intuitive insight. AI-assisted synthesis with your critical edits.

Deliverable 3 — One-Page PRD

A concise PRD for the chosen feature covering: problem statement, proposed solution, success metrics (SMART), functional requirements (user stories), and out-of-scope items. Generated with your PRD prompt from the prompt library, refined with at least one QA loop iteration.

Deliverable 4 — Prioritised Backlog

A RICE-scored backlog of at least 5 user stories or features related to your scenario. Include the RICE calculation for each, a final ranked list, and a one-paragraph rationale for the top-ranked item. Scoring done with AI, validated by you.

Deliverable 5 — AI Governance Brief

A governance brief for any AI-powered aspect of your feature (or for the AI tools you used during PM work). Must address: data and PII considerations, escalation design (if customer-facing), accuracy monitoring plan, and AI disclosure approach. One page maximum.

Deliverable 6 — PM OS Summary

A reflection document covering: which PM OS components you used during this capstone, what worked well, what you would do differently, and your personal plan for maintaining your PM OS over the next 90 days.

Capstone Rubric

Criterion	Excellent (4)	Proficient (3)	Developing (2)	Beginning (1)
AI as partner	AI used for all 6 deliverables with clear evidence of QA loops and human editing	AI used for 4–5 deliverables with some evidence of human review	AI used for 2–3 deliverables; limited evidence of critical review	AI used minimally or outputs appear unedited
Context quality	context.md is complete, concise, accurate, and would produce correct AI outputs	context.md covers all 5 sections but some gaps or inaccuracies	context.md covers 3–4 sections with notable gaps	context.md is incomplete or inaccurate
PRD rigour	PRD is clear, scoped, has SMART metrics, and shows QA loop iteration	PRD is mostly complete with minor gaps in metrics or scope	PRD structure present but metrics are vague or out-of-scope missing	PRD is incomplete or does not follow the template
RICE scoring	All 5+ items scored with justification; ranking is defensible	All items scored; some justifications thin	Items scored but RICE components poorly estimated	RICE applied incorrectly or not at all
Governance	Governance brief is thorough, specific, and cites relevant regulations	Governance brief covers all 4 areas with appropriate depth	Governance brief present but missing 1–2 areas	Governance brief missing or only superficial
PM OS reflection	Reflection is honest, specific, and includes a credible 90-day plan	Reflection covers all 4 areas with good specificity	Reflection present but vague or generic	Reflection missing or very brief

Scoring guide: 20–24 = Distinction · 15–19 = Merit · 10–14 = Pass · below 10 = Needs resubmission

Capstone Submission Checklist

Before considering your capstone complete, verify every item below.

context.md

Under 400 words · covers vision, users, problems, metrics, constraints · verified by pasting into a fresh AI chat and confirming it produces accurate product-aware responses.

Discovery synthesis

Based on ≥3 interviews (real or simulated) · pain points ranked · HMW statement present · at least one surprising insight called out.

One-page PRD

Generated using your prompt library PRD prompt · at least one QA loop iteration documented · SMART metrics present · out-of-scope section included.

RICE backlog

≥5 items · each with Reach, Impact, Confidence, Effort scores and calculation · final ranked list · top item justified in one paragraph.

Governance brief

Covers data/PII, escalation, accuracy monitoring, and AI disclosure · under one page · specific to your chosen scenario.

PM OS reflection

Names specific PM OS components used · honest about what didn't work · includes a concrete 90-day maintenance plan.

Final Verification

All 6 deliverables exist as separate, readable documents (Markdown, Google Doc, Notion, or similar)
At least 3 deliverables show clear evidence of AI assistance (e.g., note at bottom: "Generated with [tool], reviewed and edited by [name]")
At least 2 deliverables show evidence of a QA loop — a second AI pass reviewing the first output
The context.md file would allow a new team member to get productive in under 10 minutes
The governance brief is specific to your scenario — not a generic template copy

Done when: You could present these 6 documents to a senior PM or product leader and confidently explain the AI-native approach you used, the decisions you made, and how you ensured quality throughout.

Module 12 Quiz — Course Synthesis

8 questions spanning the full course · click an option to answer

1. What is the single most important file in a PM Operating System?

The context file is the foundation. Without accurate context, even perfect prompts produce generic outputs. All other PM OS components depend on context being well-maintained.

2. What is the RICE formula?

RICE = (Reach × Impact × Confidence) ÷ Effort. Reach is users per period, Impact is a 0.25–3 scale, Confidence is a percentage, and Effort is person-months. The result is a normalised score for comparison.

3. What is the key difference between AI-assisted and AI-native product management?

The core distinction: AI-assisted is reactive (use AI when convenient), AI-native is structural (AI is designed into how you work and what you build). AI-native PMs build systems; AI-assisted PMs use tools.

4. A properly written experiment hypothesis should include which three components?

The three-part hypothesis format — "If [change], then [outcome] because [rationale]" — forces clarity on what you're testing, what you expect to happen, and why you believe that. The rationale is the most commonly omitted and most valuable component.

5. In RAG architecture, what does the PM most directly control?

PMs own the document layer. This is their highest-leverage contribution to AI system quality. The quality of product specs, support docs, and policies directly determines the quality of every AI-generated answer.

6. What is a QA loop in AI-native PM work?

A QA loop is when you use a second AI prompt to review the first output — e.g., "Review the PRD you just generated. What is missing? What assumptions are made? What could be misinterpreted?" This is one of the most powerful habits of AI-native PMs.

7. What does "Now / Next / Later" represent in roadmapping?

Now/Next/Later is a time-horizon roadmap format that communicates strategic direction without false date precision. "Now" = in-progress, "Next" = committed next bets, "Later" = directional vision. It reduces the calendar commitment pressure while maintaining stakeholder alignment.

8. Which of these best describes the role of a PM in an AI-native organisation?

AI amplifies PM output but does not replace PM judgment. The AI-native PM uses AI to increase speed and quality while remaining accountable for every decision, maintaining critical thinking, and applying human context that AI cannot access.

Score: 0 / 8

Congratulations

You have completed the AI-Native Product Management course. You now have the frameworks, habits, and systems to operate as an AI-native PM — not just using AI tools, but thinking, working, and building in an AI-native way.

What to do next:

Review the Appendix for quick-reference cheat sheets and the full prompt library
Share your capstone with a peer PM for feedback
Set a calendar reminder in 30 days to review and update your PM OS
Join the Ross Thomas PM community to connect with other AI-native PMs