NIL Intelligence Licensing

You are not being replaced. You are being published.

That sentence is the whole argument. Everything that follows is the machinery that makes it true: a proposed licensing model — Name, Intelligence, Licensing — that lets experienced trust professionals keep their name on their judgment, get paid when that judgment is used, and keep getting paid after they leave.

This is a concept paper, not a rate card. The framework is what we are proposing. The open numbers are named as open.

The problem this solves

Trust officers, portfolio administrators, and senior operations staff are not resisting AI agents because they distrust the technology. They are resisting because they understand the economics better than anyone gives them credit for.

Thirty years of judgment — which discretionary distributions get approved, which fee waivers are defensible, which account will blow up in eighteen months — is the only asset a career practitioner truly owns. Today, when an agent is deployed alongside that person, their judgment is absorbed into the system for free. The institution keeps the asset. The person keeps a salary that ends the day they leave.

Framed that way, refusal is rational. Any adoption strategy that does not address it is negotiating against a wall.

NIL Intelligence Licensing removes the wall by changing what the practitioner is being asked to do. They are not being asked to train their replacement. They are being asked to publish.

The concept

College athletics solved a structurally identical problem. An athlete's name, image, and likeness had commercial value that the institution captured entirely; NIL licensing returned that value to the individual without dismantling the institution. The athlete still plays for the school. They simply own their own name.

We apply the same logic to expertise. Name, Intelligence, Licensing:

• A named practitioner is paired with an agent inside the fiduciary system.
• That agent carries the practitioner's signature — their judgment, their patterns, their standards, attributed to them by name.
• Every time the system draws on that intelligence to reach a decision, a royalty accrues.
• The royalty splits evenly: half to the named practitioner, half to the system that hosts, maintains, audits, and defends the intelligence.
• The license belongs to the person, not the employer. It travels with them.

That last point is the one that changes the conversation. A trust officer can leave the bank, retire, or move firms, and the royalty stream continues — because what is being paid for is the intelligence, not the employment.

Why an even split

It is legible, and it is defensible in both directions. The practitioner supplied the judgment; without it there is nothing to license. The system supplied the distribution, the audit trail, the regulatory posture, and the ongoing maintenance that keeps the intelligence current and usable; without that, the judgment reaches no one and earns nothing. Neither half is decoration. An even split says so without requiring a negotiation.

How the royalty works

Accrual is per decision. The royalty triggers on each decision the agent participates in — not on seats, not on a flat subscription, not on a vague usage pool. Per-decision accrual is what makes the model honest: the practitioner is paid for work actually done, and the institution pays for value actually received.

Settlement is monthly. Decisions accumulate over a calendar month and settle in a single payment. Metered like a utility, paid like a paycheck. This is deliberately unremarkable — nobody should have to think about the plumbing.

The curve starts small. Early months will be small. That should be stated plainly rather than dressed up, because a practitioner who is told to expect a rounding error in month one and receives one will trust the second statement. The value is in accumulation: every decision adds to the base, and unlike salary the base does not reset when the practitioner changes employers or stops working.

Tenure stops being seniority and starts being a catalogue.

The credit-split framework

Most fiduciary decisions of any consequence involve more than one person's judgment. A discretionary distribution gets framed by the officer, checked against precedent by someone who has seen the family for twenty years, and pressure-tested by a third party who has watched this exact pattern fail before. If the royalty model cannot attribute across those contributions cleanly, it collapses in the first committee meeting.

The answer is a writing credit.

Music publishing solved this problem a century ago, and it has held up under litigation, across catalogues, and through complete changes in how music is distributed. Every work has a credit sheet. Shares are declared at the time of creation, not argued about after the royalties arrive. The framework is legible to anyone who has ever looked at the back of an album sleeve.

Every decision produces a credit sheet with two positions:

If three intelligences are consulted, they divide the thirty. If none are, the primary takes the full share. The 70/30 weighting matters less than the fact that it is fixed and published in advance — the framework's value is that it removes the argument, not that it produces a metaphysically perfect number.

This also mirrors how a trust committee already reasons. Committees have always known who owned a call and who was consulted on it; the credit sheet simply writes down what the minutes already imply.

The hard case: when the contribution is a veto

The framework above assumes contributions are additive — someone recommends, someone else refines. Fiduciary work does not behave that way. Frequently the most valuable contribution in the room is the person who says no.

A veto is not a 30% consultation. In fiduciary practice a prevented loss is generally worth more than a captured gain, and a model that pays a blocking contribution less than a supporting one is teaching exactly the wrong behavior.

The clean resolution: a decision to decline is a decision. It carries its own credit sheet, and the intelligence that framed the decline holds the primary credit on it.

This requires only that the decision record treat "declined" as a first-class outcome rather than an absence of one — which a defensible fiduciary audit trail should be doing regardless. The person who stopped the distribution is credited as the author of the stop, not as a footnote on someone else's proposal.

Adjacent cases worth settling early

• Standing precedent — an intelligence whose prior decision is cited as controlling in a later one. Suggested treatment: contributing credit, since the judgment is genuinely in use.
• Institutional policy — where the decision is fully determined by written policy and no judgment was exercised. No credit accrues; the system takes the full royalty.
• Disputed primary — two intelligences with genuine co-authorship. Split the primary credit evenly rather than inventing a tiebreak.
• Estate and succession — whether royalties pass to a beneficiary on death, and for how long. Music publishing has well-tested answers here, and they should be borrowed rather than reinvented.
• Decay — whether a licensed intelligence that has not been refreshed in several years should attract a declining share, and how that is disclosed at signing.

What the practitioner actually sees

The credit-split framework only works if it is visible. A monthly statement should read like a royalty statement, because that is what it is — the practitioner should be able to see which decisions earned, in what role, and at what rate.

Rate per decision is left open deliberately — it is the one number that has to be set against real economics rather than reasoned from first principles. Everything else on the statement is determined by the framework.

Two design notes. First, the statement is itemized by decision class rather than by individual decision: a practitioner does not want 690 line items; they want to see where their judgment is being used. Second, the declined decisions appear as their own line with primary credit — the statement should make the veto economics visible on the page, because that is where the model earns its credibility with the people most likely to resist it.

The pitch, in one line

You are not being replaced. You are being published.

The practitioner keeps their name on their work, gets paid for it while employed, and keeps getting paid after they leave. The institution keeps the judgment it was going to lose to retirement anyway — on terms the practitioner will actually agree to. The system takes its half for making that exchange possible, auditable, and portable.

Related: NIL Intelligence Licensing in the vocabulary.

Questions still open

The framework above is what we are prepared to defend. These items are not:

• Rate per decision, and whether it varies by decision class or by assets under administration.
• License term, renewal, and termination mechanics.
• Whether the 70/30 primary/contributing weighting survives contact with a real committee, or wants to be 80/20.
• Estate and succession treatment.
• Regulatory posture — how a fiduciary institution discloses that a named intelligence participated in a decision, and to whom.
• Whether institutions will accept license portability, which is the provision most likely to draw resistance from the buyer side.