A book about governance, technology and human judgment
Judgment Capital
What Boards Must Preserve When AI Does the Thinking
As machines take over more of the work through which human expertise is built, institutions face a governance problem they have no line for: they can become more capable technologically while becoming less able to judge independently.
Judgment Capital names the asset that is draining, the six ways it drains, and what a board is accountable for while it does.
Governance
Technology
Human judgment
Judgment
Capital
What Boards Must Preserve When AI Does the Thinking
Cover design provisional
The question
What happens when an institution stays accountable for decisions it is gradually becoming less able to make without the machine?
Artificial intelligence now takes part in analysis, drafting, diagnosis, recommendation and decision. The productivity gain is real. A second consequence deserves the board’s attention.
Most of what allows a person to judge well is built by practice: working imperfect information, noticing the anomaly, being wrong, and living with the consequence. If the machine does that work, the institution needs a different question — not only what can we hand over? but what must we remain able to do ourselves?
The idea
Judgment is not only a personal quality. It is an institutional asset.
Judgment capital is the accumulated organisational capacity that lets an institution form, test, retain and exercise sound independent judgment — the capacity to know when its own instruments are wrong.
It behaves exactly like capital, which is why the word is not a metaphor. It is accumulated slowly and spent quickly. It sits behind the operations rather than inside them. It appears on the face of no statement, and the enterprise runs on it regardless. Your institution holds a quantity of it this morning; you will learn its exact size on the day you reach for it.
And it is not only spendable. It is issuable. Name the right cue, give people clean and fast feedback, and judgment forms faster than the folklore of mastery suggests. Some institutions are building the bench right now while others cut it.
Every board manages financial, operational and cyber risk. None is managing the risk that the institution is quietly losing the ability to build the judgment it runs on.
Six modes
Judgment capital drains in six ways, and each one has an owner.
Four are visible in the record already. Two are named before they are felt, which is the only useful moment to name a risk.
Pipeline collapse
The intake that no longer arrives, and the specialist who has no one forming up behind them.
Succession · CHRO
Competence drift
A control function's judgment migrating into agreement with the machine — the override rate sliding toward zero, not because the machine improved but because the humans stopped forming an independent view.
Decision-quality · CRO
Inquiry groupthink
Every seat briefed to the same depth from the same fluent source, so the room stops generating alternatives and its unanimity certifies nothing.
Decision-quality · CRO
Authorship erosion
The share of the institution's output that no one on the payroll could now rebuild from scratch, tools down.
Decision-quality · CRO
Agency erosion
A correct judgment forms in someone ready to use it — and the process will not let it reach the decision.
Decision-quality · CRO
Detection blindness
The deskilled watcher who can no longer see what the machine missed. It usually arrives as a bill rather than a warning.
Both lines · CRO + CHRO
What it looks like on the sheet
The six modes sit on two owned lines — succession, which the chief HR officer owns, and decision-quality, which the chief risk officer owns — across three tiers of escalation. Every board manages the first four risks below. The fifth is the asset all four assume.
| Ref | Risk | Owner | Reviewed |
|---|---|---|---|
| 07 | Credit and market risk | CFO | Monthly |
| 08 | Cyber and information security | CISO | Quarterly |
| 09 | Third-party dependency | COO | Quarterly |
| 10 | Key person | CHRO | Annual |
| 11 | Judgment capital — the capacity to form and hold judgment | — | — |
The book
Fourteen chapters, in four parts.
A thesis-driven instrument book rather than a survey: it names an asset, tracks its drawdown through six mechanisms and three altitudes, and ends with something a committee can put on an agenda.
Part one
The Naming
What the risk is, why no one is watching it, and why waiting for proof is the negligent act rather than the rigorous one.
- —Shut It Down
- —The Proof You Are Waiting For Cannot Arrive in Time
- —The Rung Climbed Before the Loss
Part two
The Mechanism
How judgment is actually built, how automation cuts the path that builds it, and the optimist's case put at full strength before it is answered.
- —The Nose
- —The Two Ironies
- —The Optimist's Case
Part three
The Climb
The same erosion at rising altitude — the individual, the institution, the market — each chapter carrying named, on-the-record reporting.
- —The Junior Who Never Arrived
- —A Single Point of Failure Made of People
- —Detection Blindness
Part four
The Instrument
Something a board can table on Monday: a register with owners and a clock, a ladder to climb, and the forward turn the book ends on.
- —The Register
- —Running the Register
- —The Ladder
- —The Agenda Item
- —Learning to Learn
The sources are checkable at the moment of doubt: the notes sit at the foot of the page, where a sceptical professional actually reads them, and the working methods are printed in an appendix rather than held back.
Written for
Written for the people who stay accountable.
Not for the people who deploy the systems — for the people who remain answerable for the decisions afterwards. The book assumes you cannot read a model card and should not have to in order to govern this.
Audit and risk committee chairs
The one reader with both the authority to put something on a register and a reason to buy a book about it.
Board chairs and non-executive directors
Accountable for decisions the institution is becoming less able to make unaided — and holding the only clock long enough to see it.
Chief risk officers and heads of internal audit
The decision-quality line is theirs: override rates, cold-read capacity, the share of the machine's output nobody sampled.
Chief HR officers
The succession line is theirs: the intake that forms the bench, and the apprenticeship the tooling quietly shortened.
Chief executives and senior management
The operating model is where the erosion actually runs, one automation decision at a time.
Technology and AI leaders
They deploy the systems, and they inherit the question of what the institution can still do without them.
Read from the book
At a few minutes past three in the morning, the night-shift operator did not believe his own panel.
The number on the level indicator for the heavy-ends column had been holding steady for the better part of an hour. A steady number in the small hours was, in most weeks, the most reassuring thing a control room could give you. But the operator had been on that unit for nineteen years, and a steady number on that particular gauge was the one reading he had learned not to trust. He had watched it sit flat once before while the column behind it filled past where it should have, and the flat reading had been the gauge giving up rather than the level holding.
So he made the call the procedures gave him no instruction to make, because no procedure anticipates the moment a man decides a working instrument is lying. He shut it down. He tripped the unit, took the loss in throughput, and woke the shift supervisor to explain why. The gauge had failed. He had been right, and could not have told you with certainty why.
From Chapter One · Shut It Down
The book opens in a control room because that is where the asset is visible: a man who distrusted one instrument under one set of conditions, having been fooled by it once.
He retires, and the institution replaces him — conscientiously — with a better dashboard and a checklist written to stand in for him. Everyone understands the checklist is a substitute. Nobody believes it is an equal one.
The authors
Two perspectives on one governance problem.
One author has spent her career inside the regulated institution, where judgment is governed under a supervisor’s eye. The other has spent his in the business whose entire product is judgment, and whose apprenticeship is the junior work being automated first.
Oksana Sivokobilska
Independent director · governance, digital transformation and AI
Twenty-five years inside the regulated institution, where judgment is governed under a supervisor's eye. Most boards choose between a banker who cannot read a technology architecture decision and a technologist who has never carried a P&L; she is the uncommon case of both in one seat.
Alex Blumentals
Founder and entrepreneur · AI transformation enablement
Twenty-five years in businesses whose entire product is judgment — and whose apprenticeship is the junior work now being automated first. Start from judgment, and you get people who can learn in the open rather than brace for the tool — which is what makes the implementation work.
Beyond the book
The book is where the argument is set down, not where it stops.
The research
The book is built on cross-sector case work on how professional expertise forms and erodes under automation, and on interviews with the people running the functions where it is happening.
The board conversation
Both authors speak to boards, directors' institutes and executive programmes on cognitive risk, judgment capital and what oversight of it actually looks like.
Working with us
A judgment review takes the book's instruments into one institution: the register drafted against your own processes, a rung established, a board paper that names owners and thresholds.
Board papers
Material for a board-level discussion.
Two papers written to be circulated before the meeting where this first comes up, and two things to read here without asking anyone.
Board paper
The judgment capital problem
For a committee chair to circulate before the meeting: the asset, why the proof cannot arrive in time, the six ways it drains, and what to put on the agenda.
Reference sheet
Six modes of judgment erosion
The register on one sheet — two owned lines, three tiers, six modes with their owners, and the ladder that says where an institution stands.
On this site
Five questions for Monday
For the meeting where none of this exists yet — with what a good answer sounds like and what the fluent half-answer is hiding.
On this site
The framework
The vocabulary, the register, the ladder and the review — every instrument the book specifies, in one place.
Judgment Capital notes
The first edition is open.
The book is written, and the first edition is open for pre-order: a deposit now, the balance when it ships, and a date we undertake to deliver by. Your copy is numbered, and your name goes in the list of first readers at the back.
Or stay on the list: one letter when there is something worth sending — a case, a mechanism, a figure we had to correct. No sequence, no drip.
One click to confirm. Unsubscribe in every letter. We do not share the list.
Governance
Technology
Human judgment
Judgment
Capital
What Boards Must Preserve When AI Does the Thinking