The technical floor
Trace APIs, release paths, and digital assets until product intent becomes inspectable system behavior.
A field guide for the role that already arrived
AI makes the work easier to enter—and easier to get plausibly wrong.
Follow product intent from user signal to shipped behaviour across data, design, code, AI, and delivery—without outsourcing judgment or pretending to replace a specialist.
20chapters
20chapters
24original figures
435interior pages
26workbook records
The new operating surface
Follow the work, not just the document.
PMs are still accountable for discovery, judgment, prioritisation, and user understanding. But the work now runs through systems a PM can inspect and participate in more directly—with appropriate access and specialist support: APIs, data, repositories, prototypes, evals, agents, QA, and releases.
The document-heavy loop
Research→PRD→Handoff
Product intent can fade as it travels through layers of interpretation.
The book’s operating chain
Every transition can lose intent. The book offers practices for keeping each one inspectable.
Grounded in firsthand product work acrossHeatseeker·Tokopedia·Shipper·Qlip·AvidX·Noted
Inside the book
Build your technical floor.
Five connected parts take you from system basics to an AI-augmented PM practice designed for day-to-day product work.
The data floor
Read product reality through queries, event models, and experiments—without turning a metric into a story too early.
The design floor
Make ambiguity visible in flows and reusable systems, then reason carefully about what helps users move or stop.
The AI product floor
Start with solution fit, define evaluation before launch, and design calibration loops around uncertain outputs.
The operating model
Use agents, repositories, prototypes, and evidence as shared product instruments—not as shortcuts around judgment.
The five-chapter quick path
Start applying the book before page 435.
Read Chapters 1, 8, 13, 16, and 20. Use one bounded product problem and leave each chapter with an artifact you can bring into a real team conversation.
Chapter 0101
Build your technical floor
Map where you can ask, inspect, prototype, or contribute—and where specialist ownership must stay explicit.
Participation boundaryChapter 0802
Expose one real flow
Turn actors, states, exceptions, and handoffs into acceptance criteria the whole team can inspect.
Flow-to-acceptance mapChapter 1303
Define the AI quality bar
Replace demo confidence with tasks, rubrics, thresholds, failure severity, abstention, and review ownership.
Evaluation scorecardChapter 1604
Create a portable context brief
Carry objective, evidence, constraints, tools, output, and verification into an inspectable working context.
Portable context briefChapter 2005
Carry one slice through delivery
Follow the work through implementation, review, preview, rollout, measurement, and the next decision.
Ship decision recordWant the complete practice?Read all 20 chapters in order, or begin with one evidence row in Chapter 18 and carry it through the Chapters 19–20 capstone.
A runnable companion, not bonus filler
Practice the decisions, not just the vocabulary.
The public companion pairs bounded artifacts with deterministic checks. Inspect API states, query the Sakila SQLite dataset, evaluate constructed AI outputs, and adapt the same records used throughout the book.
8practice areas
26reusable records
0API keys required
- Compare API success, validation, and provider-failure states
- Run SQL and analytics checks against deterministic fixtures
- Evaluate two AI resolver versions on held-out cases
- Inspect a synthetic, labelled customer-feedback instrument
See where intent can fade between a user signal, a decision, shipped behaviour, and the next learning loop.
Turn a persuasive demo into a testable system with tasks, rubrics, gates, severity, abstention, and review ownership.
Follow the last mile through review, preview, rollout, measurement, and the decision that comes next.
Codex + Claude CodeTwo entry points into the same bounded companion.
The repository includes project skills for Codex and Claude Code. Both route into the same 20 chapter guides and 26 workbook records. What happens next still depends on the agent, its permissions, and the tools available. Selected chapters include narrow, no-key checks; none of the fixtures verifies a real product.
Open the public repository
Read before you buy
58 pages. No email. No gate.
Download the curated reading sample and decide whether the book’s argument, evidence, and working style are useful for the product problems you actually own.
58pages
01The central argumentSee why AI expands participation without transferring specialist authority.
02Multiple evidence stylesRead practical frameworks alongside bounded firsthand product work.
03The delivery boundarySee how the book distinguishes generated output from verified, shipped work.
What the book helps you do
Work closer to where product decisions become real.
01
Ask sharper questions of engineers, designers, data teams, and AI systems.
02
Turn product ambiguity into inspectable artifacts that support testing and delivery.
03
Use coding agents without confusing generated output with verified work.
04
Design evaluation and calibration practices that keep humans in control of AI product decisions.
05
Follow product intent from signal to evidence, release, and learning.
06
Participate more responsibly inside technical delivery without pretending to replace a specialist.
Who it’s for
You own the outcome. Now you need a clearer view of the system.
Written for people already in the role — a year or more of shipping features and sitting through the arguments around them. It isn’t a “how to become a PM” primer. Product designers and engineers who work alongside product teams tend to find it maps onto their side of the same work.
Practising PMs
A year or more into the role: you already run discovery and delivery, and now want to work closer to the technical, data, design, and AI artifacts around your work.
AI product managers
You need a grounded way to define quality, evaluate uncertainty, and keep humans in control.
Product leaders
You want PMs to participate more deeply without blurring specialist roles or accountability.
Choose your edition
Read it your way.
The same complete book, available direct or through Amazon.
Recommended
Direct ebook
PDF + EPUB
DRM-free files plus the 55-page workbook with 26 reusable records and access to the public runnable companion.
- Read on Kindle, tablet, phone, or laptop
- Files delivered to your email
- Directly supports the author
Amazon Kindle
Kindle edition
Buy and read inside the Kindle ecosystem, synced across your devices.
- Delivered through Amazon
- Kindle app and device support
- Marketplace pricing
Amazon print
Paperback
A 6 × 9 inch trade paperback with a 435-page interior and 24 print-ready figures.
- Black-ink interior on white paper
- Shipped by Amazon
- Made for notes and reference
Before you buy
Good questions.
Do I need to be able to code?
No. This is not a programming textbook. It gives PMs a practical starting floor: concepts and exercises for inspecting systems, asking better questions, and participating more responsibly.
What comes with the direct ebook?
The complete book in PDF and EPUB, plus the 55-page companion workbook with 26 reusable records and access to eight public practice areas that need no model API key.
Do I have to read all 435 pages in order?
No. The five-chapter quick path takes you through Chapters 1, 8, 13, 16, and 20. There is also a chapter-by-chapter path and a Chapters 18–20 capstone path.
Can I read a sample before buying?
Yes. The 58-page reading sample is a direct PDF download with no email form, account, or checkout.
Can I read the direct EPUB on Kindle?
Yes. You can send the EPUB to your Kindle library using Amazon’s Send to Kindle service.
How does email delivery work?
After Stripe confirms payment, a private download link is sent to the email used at checkout. Keep that email so you can retrieve the files later.
Is this an “AI will replace PMs” book?
No. The argument is that AI expands where product judgment can be applied. It does not transfer specialist authority or reduce the cost of being wrong.
The field guide for the role that already arrived