New from StackedCFO

FIRE
THEM ALL

A CEO's Unfiltered Takes in the Age of AI

"The org chart had eleven names on it. Eighteen months later it had one."

I spent eighteen months replacing my own corporate development team, my accounting function, and most of my own job with AI agents. This is the honest account — what worked, what it cost, and what was left when the org chart had one name on it.

266 pages 48 chapters 5 fictional interludes By Casper Zhao, CPA
Fire Them All book cover — an empty executive chair faces a city skyline, with scattered gold nameplates of eliminated executive roles on the floor in front of the desk
11 → 1
Headcount, corp dev function
$450K
Annual cost, fully replaced
7
Agent stacks now running
13yrs
Author's QoE / audit background
Section 2

Who this book is for.

Written from inside the function, not about it from a distance.

CFOs
Controllers
FP&A Leaders
Big Four Professionals
Private Equity Operators
Founders
Finance Transformation Leaders
Anyone whose job touches a spreadsheet
Section 3 — The Problem

Most finance leaders are running 2019 headcount on 2026 capability — and nobody has told the board.

Six Saturday mornings. That's how long it took to build a process that did the work of a three-person analyst bench, for under two hundred dollars a month.

Here is the operating reality most finance leaders are quietly sitting on. The close still takes ten to twelve days, run by a team sized for a workflow that no longer exists in its original form. The corporate development function still staffs a junior analyst bench to produce screens, models, and first-draft memos — work that current tools can now do, end to end, in under an hour, at a cost that rounds to nothing against the loaded cost of the team producing it. The offshore accounting relationship that was supposed to be the cost-efficient option is now, itself, the most expensive way to do bookkeeping that current technology supports. None of this is a secret inside finance organizations. It is the thing nobody on the leadership team wants to be the one to say out loud, because saying it out loud means deciding what to do about it, and deciding what to do about it means having a conversation with real people about real jobs that most executives would rather defer.

That deferral has a cost, and the cost compounds. Every quarter a function runs on its old headcount while the capability gap widens is a quarter of margin given away to organizations that made the harder call earlier, and a quarter closer to a board member, an investor, or a competitor asking the question directly instead of waiting for leadership to raise it first.

"The gap between what AI can already do and what organizations have actually deployed isn't a technology gap. It's a decision gap. This book is what it looks like to close it — and what it costs."

I wrote the memo that would have ended four careers in one move. I almost sent it. This book is about the eighteen months between that memo and the org chart I actually run today — one name at the top, seven agent stacks underneath, and a much harder answer to the question of what was actually worth keeping.

This isn't a vendor pitch and it isn't a doom book. It's a CEO's actual ledger: the corporate development team that trained its own replacement, the executive assistant function that runs on five coordinated agents, the offshore accounting team I should have told sooner, the financial advisor relationship worth keeping and the one that wasn't, and the six weeks I spent trying to automate my own job — and why I stopped.

Section 4 — Spicy Snippet

Nine nameplates. One floor.

Every role on the cover is a real chapter in the book. Hover or tap a plate to see what happened to it — and where to read the full account.

↳ nobody asks what happened to the people. click any plate.
Eliminated, Month 3
CHIEF STRATEGY OFFICER
Eliminated, Month 11
VP OF OPERATIONS
Eliminated, Month 6
VP OF MARKETING
Eliminated, Month 9
CHIEF FINANCIAL OFFICER
Attempted, Not Cut
CHIEF INFORMATION OFFICER
Eliminated, Month 4
HEAD OF GROWTH
Eliminated, Month 7
HEAD OF ESG
Eliminated, Month 5
CHIEF PEOPLE OFFICER
Reclassified, Not Cut
HEAD OF ANALYTICS
From Chapter 1 — The Number at the Top of the Page

Vera asked the question four seconds into the meeting, before I had finished my second sentence, and I have replayed those four seconds more times than I have replayed almost anything else in my career.

I had built two slides. The first showed our corporate development function's deal volume for the trailing twelve months. The second showed the same volume, produced by an AI-assisted process I had built over six weekends, at a fraction of the cost. I had practiced the transition between the two slides. I had not practiced what came after, because I had assumed, with the specific confidence of someone who has just spent six weekends being very impressed with himself, that the slides would speak for themselves.

Vera looked at the second slide for four seconds. Then she asked: "Which of these deals actually closed because of something a person did, not the model?"

I did not have an answer. I want to be precise about what that means, because it is not the same as having a bad answer. I had built two slides comparing cost and output volume, and I had not built — had not even thought to build — the slide that would have answered the only question that mattered, which was not how much cheaper the work was, but how much of the work that mattered had ever been the work I was comparing.

This book exists because of those four seconds. Everything that follows is the slow, sometimes embarrassing, occasionally costly process of building the answer I did not have in that room.

Continue reading in Chapter 1 →
Section 5 — Secret Sauce

Ten reasons this isn't another AI book.

What differentiates this from the rest of the AI-and-work shelf:

PART I
13 chaptersThe Team
A corporate development function watches its own workflow get absorbed into a system it helped build. Includes the Klarna chapter, the deal that nearly died over something no model contained, and the vendor demo that almost cost four careers before anyone tested it on real data.
PART II
11 chaptersThe Accountant
What AI can and cannot do with your books, your taxes, and the letter from the IRS. A full walkthrough of a real CP2000 notice, the five things you must document before you file, and the related-party transactions that will confidently get misclassified.
PART III
10 chaptersThe Advisor
The one-percent-a-year fee, what it actually buys, and what a model now does for free — including the annuity that quietly cost a client $240,000 she didn't know she was spending, and the advisor whose practice got better by being transparent about the tools.
PART IV
9 chaptersThe Prompts
The unglamorous, high-return prompt library that actually gets used: the monthly variance review, the contract red-flag scan, the what-am-I-missing prompt — and the one you should be most afraid of, the obsolescence audit you can run on your own job tonight.
PART V
7 chaptersThe Agents
The book the chart demanded. The org chart with one name on it, function by function: the EA stack, the FP&A stack, the offshore accounting transition, the diligence stack eighteen months later — and the six weeks spent trying to build a CEO agent, and why it stopped.
From the Pages

Ten lines that will follow you out of the book.

No context. No setup. Just the sentences readers keep coming back to.

I had built two slides comparing cost and output volume, and I had not built the slide that would have answered the only question that mattered.

Chapter 1

The agent has nothing to lose. That sentence is the shortest, most complete explanation of why the CEO function has not been automated.

Chapter 47

Someone is going to run this analysis on your team. The only decision you have is whether that person is you.

Chapter 13

Gratitude from the trusty toward the warden is stranger and more real.

Interlude III — The Raise

The relief told me I had found the ceiling. What I had actually found was that the ceiling was higher than I thought.

Chapter 2

The twenty percent is not twenty percent of the hours. It is twenty percent of the hours and one hundred percent of the value.

Chapter 4

I had automated the homework. I had not gone anywhere near the exam.

Chapter 2

Catching it just proved the model needs exactly one human babysitter — which is why they alone were kept, which is the whole reason they can never leave.

Interlude II — The Flux Review

The most honest meeting about the firm's economics that anyone could remember having.

Chapter 9

What is left is shrinking. Not toward zero, but toward a smaller number than it was eighteen months ago — and the rate is accelerating.

Chapter 41 — What's Actually Left
Section 6 — Why This Matters Now

The capability is already here. The deployment isn't. That gap is the opportunity — and the risk.

Independent research on AI task coverage by occupational category shows a consistent pattern: theoretical AI capability in business, finance, legal, and administrative work already sits in the 0.8–1.0 range on a 0–1 scale. Observed, actual deployment in those same categories sits closer to 0.1–0.2. That gap exists in every finance organization currently running pre-2023 headcount against 2026-grade tooling — which, at this point, is most of them.

Three forces are closing that gap faster than most leadership teams are prepared for. Private equity operators are demanding AI-enabled finance functions as a condition of platform investment, because the cost structure of the old model no longer clears their return thresholds. The agent economy has matured past single-task copilots into orchestrated, multi-step workflows — sourcing, screening, modeling, and reporting handled end to end with a human review gate, not a human doing the work with AI assistance. And corporate finance itself is evolving: the FP&A analyst, the deal team associate, and the bookkeeper are being redefined around judgment and accountability rather than production, whether or not the org chart has caught up yet.

The organizations that move first on this aren't doing it recklessly. They're doing it the way this book documents: function by function, with a named human accountable at every review gate, and an honest account of what gets lost along the way. The organizations that wait are not avoiding the decision. They are deferring it to a board member, an investor, or a competitor who will eventually make it for them.

Theoretical AI coverage — Business & Finance
~0.85
Observed deployment — Business & Finance
~0.15
Theoretical AI coverage — Office & Admin
~0.90
Time to close that gap, this book's argument
18 months
Between Every Part

Five short fictions. Two people. One layoff, told from both sides.

Interlude I
The Tuesday
Her badge stops working two minutes before her alarm. By noon she's at the Arboretum, and the relief confuses even her.
Interlude II
The Flux Review
11 PM, alone on the floor, he catches what the model missed — and understands exactly why that's the reason he's still employed.
Interlude III
The Raise
The retention bonus, funded by the cuts, closes his daughter's tuition gap. The gratitude is real. That's what's unsettling.
Interlude IV
The Invite
Two calendars, two "knowledge transfer" invites, eight days out, eleven hundred miles apart. Neither one has replied yet.
Interlude V
The Pitch
She's the one pitching the displacement engagement now. A twenty-six-year-old in the twelfth room asks the question nobody else did.
CZ
Section 8 — About the Author
Casper Zhao
Founder, StackedCFO LLC

Casper Zhao is a CPA and finance operator whose experience spans Big Four advisory, capital markets, controllership, technical accounting, FP&A, and private-equity-backed businesses. He spent thirteen years at PwC, BDO, and CFGI conducting Quality of Earnings analyses on more than one hundred transactions, and holds an active CPA license in Massachusetts.

His work focuses on building AI-enabled finance operating systems, transaction readiness programs, forecasting infrastructure, and modern finance organizations — the same systems described in this book.

He is the founder of StackedCFO LLC, a nationally available fractional CFO practice, and lives in Boston.

Section 9 — Need More Than A Book?

Need more than a book?

StackedCFO works with founders, CFOs, controllers, investors, and operators to build institutional-grade finance systems — the agent stacks this book describes, built for your organization.

FP&A Technical Accounting Board Reporting Transaction Readiness Finance AI Enablement Fractional CFO Services
Section 10 — The Operator Brief

The Operator Brief: From inside the data room.

Weekly insights on AI and finance, FP&A, corporate finance, technical accounting, and private equity — written the same way this book is, from inside the function.

AI & Finance FP&A Corporate Finance Technical Accounting Private Equity
Section 11 — Frequently Asked Questions

Questions, answered directly.

Yes. The book is written from the vantage point of a CFO and former Big Four QoE lead, and Parts I, II, and V deal directly with corporate development, accounting, and FP&A functions that CFOs and Controllers run day to day.
Yes. Part I covers diligence and corporate development specifically, including a chapter on what happens to deal quality when AI-driven sourcing and screening replace a junior deal team — directly relevant to PE-backed platforms running roll-up strategies.
Yes. Part IV's prompt library and Part V's agent-stack breakdowns are written to be usable by any operator, not just trained accountants, and each chapter explains the underlying concept before getting technical.
No. Technical terms like ASC 805 and QoE are explained in context. Readers with accounting backgrounds will get additional value from the precision; readers without one will not be lost.
The book argues it already is, selectively, and that the gap between what AI can do and what most organizations have deployed is a decision gap, not a capability gap. It also documents, specifically, what has not been automated and why — including a full chapter on the author's own attempt to automate his own CEO function.
No. The book documents both the cost savings and the human cost in the same chapters, including a chapter on how to handle an offshore vendor transition honestly, and a full chapter on severance negotiation for anyone on the receiving end of one of these decisions.
Yes. Part IV is a full prompt library covering monthly variance review, contract red-flag scanning, and an obsolescence audit prompt readers can run on their own role. The appendix consolidates the highest-value prompts from across the book.
266 pages, 48 chapters across 5 parts plus 5 fictional interludes. Available in Kindle and paperback.
Fire Them All is the operating ledger — specific functions, specific agent stacks, specific dollar figures, told as a CEO's first-person account. Related titles like The CFO Is Dead and Do You Actually Need a CFO? approach adjacent questions from different angles. See "Continue Reading" above.
Yes — an excerpt from Chapter 1 is included above on this page, and Amazon's "Look Inside" feature on the book listing includes the full first chapter.
Yes. StackedCFO builds the same agent stacks described in the book for client organizations — FP&A, technical accounting, board reporting, transaction readiness, and fractional CFO services. See the consultation section above.

Someone is going to run
the numbers on your team.

Make sure it's you. Make sure it's now.

Available in Kindle and Paperback · 266 pages · stackedcfo.com · Contact · About