Author's Preface

I. In 2026, Everyone Is Asking the Same Question

In 2026, the FDE — Forward Deployed Engineer — is suddenly everywhere: the consulting giants have turned it into a flagship service line, the model companies are stepping in to build delivery organizations of their own, and job boards keep adding postings with the exact title week after week.

Beneath the noise, I kept hearing the same set of questions — and the people asking them were getting easier to name:

• Engineers ask: how is this different from outsourcing, from on-site consulting? Will I be the one told to transfer into this role?
• Tech leads ask: we hired AI engineers and the projects still won't land — should I be hiring an FDE instead?
• Customers ask: the vendor says "it's working great" — how do I actually know when it's ready to go live?
• Founders ask: how do the economics of this business actually work?
• Skeptics ask (and with reason): isn't this just consulting wearing a new name?

Behind all these questions sits the same more fundamental thing: the stretch of road between an AI demo and customer value — which bridges need building, and who builds them. That is what this book answers. It does not answer "is FDE a fad" — that's a subjective question. It answers "what is an FDE, how do you do the job, how do the numbers work, what happens to the people" — the practical, operational questions.

II. Why I'm the One Writing This Book

I'm an AI entrepreneur myself, and I've done a number of enterprise AI projects — at the start, feeling my way forward as I went, like everyone else.

Plenty of articles have been written about FDE, but books that treat it systematically are rare. And on this topic, a great deal of the material has already warped through round after round of retelling — every link in the chain simplifies and extrapolates, and by the time it reaches the last link, the truth has gone as blurry as a photocopy of a photocopy, ten generations deep.

So I wrote this book from my own project experience and the experiences of friends around me, combined with community discussions, public materials, and a number of excellent open-source projects — every cited source is marked in the body text and the appendices.

The cases in this book come from three kinds of channels: my own delivery and observation experience; first-hand accounts from people I know personally (all used with consent, and de-identified); and public materials (interviews, podcasts, community discussions, company announcements, open-source projects, and more — see Appendix C).

III. How to Read This Book

The book is six parts, and every chapter answers one question — and only one:

1. Part One · What an FDE Actually Is — what the role really is, who it resembles and who it doesn't, whether it's a passing wave, whether AI will replace it.
2. Part Two · How the Problem Gets Solved — once you're on site at the customer: how to pick the problem, how to build a minimum viable deployment, how to answer "when can this go live."
3. Part Three · Engineering Foundations — data and permissions, constraining agent actions, evaluation, proving yourself, cost — the eight walls between PoC and production, climbed one at a time.
4. Part Four · The Real Work Begins After Go-Live — why nobody uses the system after it ships, why this has to be an executive-sponsored project, how to hand the system over so the customer can run it themselves, and why China plays by a different set of rules.
5. Part Five · How Does This Business Actually Work? — how to count the big ledger, how to charge, why customization must decline, how field experience becomes product.
6. Part Six · People and Organization — the FDE's day, the capability model, making the transition, building the team, going solo — and the role's endgame.

Each chapter's primary audience is marked right in the briefing at its head — engineers, tech leads, customers, founders, skeptics: each will find their own chain of questions laid out in order in the table of contents.

Readers with time to spare should start at Chapter 1 and read straight through. Readers in a hurry can enter at Part Two or Part Three — the engineering chapters stand on their own. Readers who just need to make a decision can go directly to Part Five for the economics, then circle back to Part One.

IV. Before We Begin

What this book has to say really comes down to one line: do the things that don't scale — at scale. The first half is the FDE's daily reality: every customer engagement is one-of-a-kind, unrepeatable grunt work. The second half is where the craft's value lives: distill what can be repeated out of that grunt work into product, so the next customer gets there faster.

The name FDE may change; the responsibility won't. If you take away only one thing from this book, I hope it is this one.

One last, honest thing: at most, this book can flag the potholes I've already fallen into, so you skip a few steps of wasted motion. Real judgment still has to be forged on-site, at the customer, bruise by bruise. However closely you read, it's no substitute for stepping onto the field yourself.

You're welcome to follow me on X (Twitter) at @richard_xu_001, or come browse my personal site at 0to1.site, where I keep sharing what I'm seeing and learning.

So begins the book.

Richard Xu

September 2026 Hongqiao, Shanghai