Product & AI Inside the book

Read a complete chapter

Which AI idea
should you build first?

Chapter 2 · Picking Your Battles

The CEO wants “AI everywhere by Q3.” Dana, his VP of Product, needs a first move she can defend.

Follow the team from a wall of competing ideas to one chosen workflow. Leave with a way to decide what to build, what to queue, and what to park.

Explore the chapters

14 original pages · Printed pages 15–28

An open-book illustration of Chapter 2, pages 22 and 23: The Selection Meeting and its workflow selection scorecard.
Chapter 2 · The Selection Meeting · Pages 22–23

How the book teaches

A story you can follow.
A discipline you can use.

Across one year, Meridian — a fictional benefits-software company — learns to build and operate AI. Each chapter pairs the story with practical frameworks, working artifacts, and five things to do Monday morning.

The chapter guide

One year. Thirteen decisions.

Read in order, or expand a chapter to find the discipline your team needs next.

Part I

The Reframe

Choose the work before choosing the technology.

01Core AI Literacyp. 1

Understand what language models do, how they fail, and which controls your team owns.

02Picking Your Battlesp. 15

Turn an AI wishlist into a ranked portfolio, with one first workflow and clear reasons for what waits.

03Model Choice Is a Posture, Not a Pickp. 29

Compare models against your constraints, record the evidence, and plan for a future change.

Part II

The Control Levers

Specify behavior. Prove it. Keep it under control.

04Prompts Are Specsp. 43

Replace a vague prompt with a versioned contract: boundaries, evidence, refusal paths, and an output schema.

05Evaluationp. 61

Build a golden dataset, separate release gates from softer scores, and measure behavior over time.

06Prompt Opsp. 77

Give changes owners, staging environments, approval rules, and a rollback you can rehearse.

07Groundingp. 89

Connect answers to knowledge with owners, effective dates, access rules, and measurable retrieval quality.

Part III

The System

Design the product around the model.

08Architecture & Integrationp. 107

Build for latency, outages, exceptions, and the trace that explains what happened.

09The Human Interfacep. 121

Design for real human attention: show uncertainty, prevent rubber-stamping, and learn from corrections.

10Tools, Agents & the Autonomy Ladderp. 139

Put actions behind enforceable boundaries and earn each increase in autonomy with evidence.

Part IV

The Operating Envelope

Make the discipline sustainable.

11Enterprise Risk, Ethics & Compliancep. 157

Turn governance into controls and evidence your team can produce when a customer asks.

12Economics of AIp. 173

Connect cost per run to business outcomes and build an argument finance can examine.

13The Operating Playbookp. 187

Bring the work together in a repeatable cadence, clear ownership, and a one-page operating standard.

Plus the QuickStart, six appendices, and a companion library of templates.

Put it into practice

Take the ideas into your work.

The companion library holds the templates behind the discipline.

Explore the Library

Your next chapter

Make the next decision
a better one.

A practical discipline to bring back to your team.

Paperback · Hardcover · Kindle

Find your edition on amazon.com

Inside Product & AI

Chapter 2 · Picking Your Battles

Page 15 · 1 of 14

Original page 15 of Product & AI, Picking Your Battles.
Open page at full size ↗Download chapter PDF

Product & AI

Choose your edition.

Book format

English-language edition · Prices in USD.

Buy paperback on Amazon

Formats vary by Amazon store.