← All books
The Decision Layer · Volume One

The Decision Gap

Why you can trace every step and still not know why

JamalNalyqorFirst Edition, August 2026

Every team running AI in production can replay what their system did. Almost none can explain why it did it. This book is about the gap between those two facts, what it costs, and why no dashboard closes it.

Execution is observable. Decisions are not. Traces, evals and dashboards were built to answer “what happened,” and they answer it well. The question that eats the week is “why did it choose that,” and it turns out none of the usual instruments were pointed at it. Volume One names that gap, shows how it hides inside clean traces and passing evals, and follows the money to where it quietly leaks.

It is written for the people who own an AI system in production: the CTO who has to defend it, the engineer who has to debug it, and the operator who has to trust it. No framework, no vendor pitch. One argument, worked through.

What's inside

Three things you'll be able to explain afterwards.

  • Why full traces still leave you guessing
  • What a silent failure actually looks like, stage by stage
  • Where the money goes when nobody is measuring decision quality
Contents

Four parts, eight chapters.

    • Introduction: The Week You Lose to "Why"
  2. I

    The Shape of the Gap

    • "We Have Observability" Is a False Sense of Safety
    • Decisions, Not Execution
  3. II

    Why Your Instruments Can't See It

    • Passing Every Eval, Failing in Production
  4. III

    The Failure That Never Announces Itself

    • The Anatomy of a Silent Failure
  5. IV

    What the Gap Costs

    • The Hidden AI Tax
    • The Economics of Reasoning
    • The Trust Gap
From the introduction

I want to start with the week that taught me what this book is about.

An agent I was responsible for had started calling the right tools in the wrong order. Not always. Only under certain combinations of context and conversation history, which is another way of saying: only when it mattered.

I did what you do. I opened the trace. The trace was clean.

The Decision Gap, page 1
About the author

Jamal Jackson is a fractional CTO and the founder of FAI, where he builds production AI systems for companies that need them to hold up. He has been shipping software since 2008 and writing about the decision layer in public since the problem first cost him a week.

More about Jamal →