A note from the founder

A Note from the Founder

Most firms that buy AI get exactly what they paid for: a system that works. What they do not get is a system they can stand behind.

That gap is where the money goes. Firms lose it in refunds, in write-offs, in senior hours spent redoing finished work, and in growth that stalls because every new client means new hires. The AI did not fail. The investment just never reached the operations where the returns are.

When the gap closes, the same spend starts paying. Work goes out right the first time. Questions that took weeks to answer take minutes. The team you have handles the growth you win. That is what AI was supposed to buy, and it is available to any firm willing to run five steps in order. This book is those five steps. It is not a product pitch. The discipline works regardless of who builds it.

But I will be direct. We built PitCrew because we believed this discipline needed a machine behind it, not just a book about it. If you want to see what it looks like running on real accounts, chapter ten tells you how.

Rishi Kulkarni

Co-founder & CEO, PitCrew

Why this book, and why now

Two ways the same AI spend behaves.

The numbers in the brief split firms into two groups. Which group you are in depends on how the spend behaves.

Ungoverned AI

An expense

Tools at the edges, demos in the middle

Output nobody trusts with real operations

The cost base it was bought to change, unchanged

Renewed every year. Returns more demos

Governed AI

An investment

The same models, wrapped in your firm’s own rules

Tested against your own history before it runs

Work you can bill, defend and build on

Returns collected every quarter, on the operating line

The return shows up in four numbers: cost per unit of work, errors reaching clients, time to answer an audit, and volume handled per head. Chapter two prices each one.

Who this book is for

Four signs this book was written for your firm.

An error reaches a client as money.

A wrong fee, a wrong payment, a wrong claim. When mistakes are denominated in dollars, the output has to be provable, not plausible.

Someone can ask “how do you know?”

A board, a regulator, an auditor, an acquirer, a client. If that question has to survive scrutiny, a confidence score will not carry it.

The rules live in people’s heads.

Approvals, exceptions and definitions of correct that have never been written down anywhere a machine could read them.

Growth currently means hiring.

Every new account or client adds headcount, because every process is checked and approved by hand.

A NOTE ON INDUSTRIES

Wealth management or logistics, insurance or healthcare, banking or the enterprise back office: the operations differ, the failure mode does not. Wherever an output must be defensible, a fee, a payment, a claim, a contract, a compliance filing, the five steps apply unchanged.

Method and sources

Where the numbers in this book come from.

A book about provable output should say where its own numbers came from. Each one below is research, estimate, or illustration.

95% of AI investments never reach production.Published research. Quoted as reported.	MIT NANDA, The GenAI Divide, 2025
AI costs have fallen close to 300x in two years.	Public model pricing, 2024 to 2026
The time figures on the ten operations pages.Marked INDUSTRY ESTIMATE wherever they appear. Ranges, not averages.	Practitioner interviews and firm process reviews
Fee errors cost advisory firms $50K to $150K a year.A range reported across operations leaders at mid-size firms. Treat it as an order of magnitude, not a benchmark.	Industry estimate
Accounts, names, ledger balances and rule numbers.Composites. Every example in this book is constructed. No client, firm or transaction is real.	Illustration
Half of CEOs cite linking AI to the P&L as a key barrier; 14% have defined that impact.Published survey research. Quoted as reported.	BCG AI Transformation CEO Survey, 2026

What this book is not

A benchmark study. We did not measure firms against each other, and we make no claim about how many have adopted any of this.

Corrections

If a figure here is wrong, we would rather know. Write to sales@gopitcrew.com and the next edition will carry the correction.