Consulting has run on the same machine for fifty years. People bill hours. Hours produce decks. Decks justify fees. Promotion, pricing, and partner compensation all orbit one number: utilization. The model worked because the scarce input was qualified human time, and human time was expensive to assemble and hard to copy.
AI removes the scarcity. When a system can produce in forty minutes what a team produced in forty hours, the input that the whole model priced has lost its scarcity premium. This is not a productivity story about doing the same work faster. It is a repricing event. Three shifts follow from it, and they reinforce each other: how firms are paid, what they measure, and what they actually hand over.
From Fee-for-Service to Fee-plus-Upside
Hourly billing is a way of selling effort. It survives only as long as effort is the bottleneck. Once execution collapses toward zero marginal cost, charging by the hour becomes indefensible, and not for ethical reasons. The arithmetic stops working. A client who knows a model drafted the analysis will not pay forty hours of rates for forty minutes of compute, and a firm that pads the timeline to protect the hours is running a race toward irrelevance.
What fills the gap is skin in the game. The next generation of advisory firms will look less like billing shops and more like micro private-equity operators. They will take a fee to cover baseline cost, then layer in warrants, success payments, and outcome equity. The pitch shifts from "we will advise you" to "we will co-own the result." That single change reorders who is worth hiring. Slide-makers lose. Capital allocators win. The advisor who can underwrite risk, structure upside, and govern execution commands a premium that day rates never could.
Ask whether your advisor would take equity in their own recommendation. If the answer is no, you are paying for confidence they do not have.
The honest objection is that most firms lack the balance sheet and the risk appetite for this. Private-equity economics require portfolio diversification that a two-hundred-person firm cannot replicate, and one bad outcome can swamp a thin book. That is true, and it is a real constraint. But the claim was never that every firm becomes a buyout fund. It is narrower and harder to dodge: the firms that figure out selective upside participation, on engagements they understand well enough to underwrite, will take share from the ones still defending the day rate. Selective is the operative word. The skill is knowing which results you are willing to co-own.
The End of Utilization
If fees move to outcomes, the internal scoreboard cannot stay anchored to hours. High utilization once signaled a well-run firm. Soon it will signal a firm that has not understood what it is selling. Keeping people busy is only a virtue when busy people are the product. When an agent absorbs the production work, an eighty-five percent utilization rate measures how much human time you are still consuming to reach a result, which is a cost, not an achievement.
The replacement metric is decision throughput per partner. Not how busy the bench is, but how many consequential decisions the firm enabled this quarter, and how good each one was. That reframing touches everything downstream. Hiring tilts toward fewer analysts and more decision architects. Pricing attaches to throughput and outcome, not effort. Comparison changes: a partner who enables fifty high-quality client decisions a quarter outranks one who keeps twelve juniors at high utilization. The uncomfortable corollary is that the firms proudest of their utilization numbers today are the ones most exposed tomorrow.
There is a fair counterpoint, and it deserves to be stated plainly. Utilization is not only a billing metric. It is also a proxy for something real: apprenticeship, mentorship hours, institutional learning, the slow transfer of judgment from senior to junior. Strip out utilization tracking without replacing those functions and you hollow out the firm's ability to make the next generation of partners. The answer is not to defend a billing metric as a training proxy. It is to build the development systems explicitly, rather than letting them ride invisibly on a number that is about to lose its meaning.
From Deliverables to Governance Products
Both shifts point at the same question: if not the deck, then what is the premium product? Nobody wanted the deck for its own sake. They wanted a system that tells them who decides what, when, on what evidence, and what happens if they are wrong. That is a governance product, and it is the part of the work that does not commoditize.
The reason is structural. AI can generate a deck in seconds, so insight stops being scarce. What remains scarce is the structure of how a specific organization makes decisions under uncertainty. Governance design means building auditable decision systems: who holds authority, what triggers escalation, what evidence threshold is required before capital is committed, and what the reversal protocol is when an assumption fails. This is bespoke, high-trust work. It cannot be templated, because it requires deep reading of one organization's risk appetite, its political dynamics, and its operational constraints. A model can draft a recommendation. It cannot, on its own, install the authority to act on one.
The commodity is not insight anymore. It is the structure of how an organization decides under uncertainty. The deck is just the receipt.
The predictable pushback is that "governance" sounds like bureaucracy, and clients want speed, not more process. Correct, if governance means adding layers. But the point is the opposite. It is replacing ad-hoc decision-making with repeatable, evidence-backed systems that accelerate the decisions that matter, because nobody has to relitigate who was allowed to decide or what counted as sufficient proof. Less process. Better architecture. Done well, governance is not the brake. It is what lets a firm move fast without breaking trust.
What Survives Handover
The three shifts are one shift seen from three angles. Outcome-linked fees force a firm to care about what actually happens after the engagement ends. Decision throughput, rather than utilization, measures whether the firm moved real decisions. Governance products are the artifact that makes both possible, because you cannot underwrite an outcome you have not built the decision machinery to govern. The deck was a record of advice given. The governance product is a working system that keeps running after the advisors leave.
This is the standard BOST works to. We are paid on consulting, on success, and on licence for the systems we hand over, not on hours logged. The test we hold ourselves to is simple and unforgiving: what survives handover. A recommendation that needs its authors in the room was never finished. A governance product that keeps deciding correctly once we are gone is the only deliverable that earns the upside.
None of this requires a capability the profession does not already have. The models exist. The pricing structures exist. The governance methods exist. What it requires is the decision to reprice the work before someone else does it first, because the deliverable era does not end on a schedule. It ends the moment a competitor stops selling effort and starts co-owning the result.