They Bought the Cover FINAL RESTORED

Michael Carroll | The One-Degree Dispatch | Critical Decision Infrastructure

They Bought the Cover. The Enterprise Paid the Bill

AI is exposing the old bargain behind consulting, where executives bought

protection for decisions the enterprise had to carry. The answer is not

cheaper advice. It is Critical Decision Infrastructure. By Michael Carroll

Research Fellow, LNS Research | Founder, The One-Degree Dispatch | Architect of Industrial Transformation

Lead image. A boardroom after the decision should already have been made. The deck is open. The invoice is aging. The evidence is present. What is missing is not information. It is ownership.

At some hour after the meeting ends, when the last laptop is closed and the room has given back its borrowed confidence, the enterprise is left with two artifacts. One is the deck. The other is the bill.

Michael Carroll | The One-Degree Dispatch

The deck has done its work. It has made the hard thing sound processed. It has turned disputed judgment into a recommendation. It has taken what people inside the business already knew and returned it to them in the language of independent authority. The problem looks cleaner now. The decision looks safer. The people who were paid to decide have something to stand behind. The bill is different. It does not sit in the boardroom. It moves into the enterprise. It becomes a capital delay, a reorganization, a plant closure, a program office, a write-off, a new operating model, a supplier change, a systems program, a cost reduction target, or a promise someone downstream must make real. The executives bought the cover. The enterprise paid the bill. That is the bargain AI is beginning to expose. The Financial Times reported that AI is forcing McKinsey and its peers to rethink pricing as clients question the old relationship between consulting effort and consulting value. If a machine can compress research, diagnostics, synthesis, benchmarking, and first-draft analysis, the client has a right to ask why the old price should survive in its old form. The issue is not only whether firms charge by the hour or by the outcome. That is the pricing symptom. The deeper issue is whether the client was ever paying only for the work. Mark Minevich put the harder point into public view in his LinkedIn post. The premium was never only for advice. It was for cover. It was for the famous name on the page, the boardroom permission, the ability to say that the best people had looked at the problem and reached the same conclusion management now wanted to defend. That does not make consulting worthless. It makes the transaction more honest. Great consultants create real value. They see patterns across companies that insiders cannot see because they are living inside them. They bring pressure to assumptions that have become customary. They can name a failure mechanism before the enterprise has language for it. They can help a leadership team face what it has delayed too long. That work matters. But the enterprise should know what it is buying. Expertise is one thing. Capacity is one thing. Perspective is one thing. Implementation help is one thing. Executive insulation is another. Too often, all of it came bundled under advisory services. The invoice did not say cover. It did not have to. Everyone understood the line item without writing it down.

The invoice said advisory services. The product often included executive insulation. The enterprise paid for both

Michael Carroll | The One-Degree Dispatch

The cover was never for the enterprise. The enterprise did not need protection from its own evidence. The cover was for the executives who needed a way to act without standing alone. The enterprise carried the cost of that protection, then carried the consequences after the protection had done its job.

The answer was often already in the room

The most revealing part of a major consulting engagement is often not the final recommendation. It is the recognition in the room when the recommendation lands. People nod before the partner finishes the sentence. They have heard this before. Sometimes they have said it before. They may have said it in operating reviews, in plant visits, in finance meetings, in customer escalations, in postmortems, in risk sessions, or in a hallway after the meeting where people speak more plainly than they do in the room. The missed shipments were visible. The margin leakage was visible. The aging capital request was visible. The system that never truly became a system was visible. The project with no real owner was visible. The plant that should have been consolidated was visible. The customer that had become too costly to serve at the promised price was visible. The executive everybody worked around was visible. No one was confused. That is why the silence mattered. Large enterprises do not always lack knowledge. They often lack a legitimate path from knowledge to action. They can see more than they can decide. They can discuss more than they can change. They can escalate more than they can resolve. The result is a market for outside authority. The consultant arrives not because the truth is absent, but because the truth has not been granted permission. The consulting process then performs a conversion. Internal evidence becomes external findings. Operational frustration becomes a diagnostic. Political conflict becomes a workstream. Delayed judgment becomes a recommendation. The enterprise receives its own knowledge back in a form the hierarchy can accept. That conversion has value when the enterprise truly needs help. It becomes costly when it substitutes for leadership. The problem is not the outsider. The problem is the internal architecture that makes the outsider necessary for decisions the company should have been able to make from its own evidence. That is why AI’s pressure on consulting pricing matters. It is not merely a story about cheaper decks. It is a story about the breakdown of a covering transaction. When AI reduces the cost of the artifact, it makes the hidden value of the cover easier to see. Once the artifact becomes cheaper, the cover has to justify itself.

Michael Carroll | The One-Degree Dispatch

What the old bundle really sold

For decades, elite consulting was sold as expertise, and some of it was. But the bundle contained far more than expertise. The client bought research, interviews, benchmarks, financial models, operational diagnostics, maturity assessments, competitive comparisons, program designs, and carefully sequenced slides. The client bought the presence of a partner who could speak with confidence in rooms where confidence itself can become evidence before the facts have been tested hard enough. Inside that bundle sat the real premium. The consulting firm provided transferred authority. It gave the executive cognitive cover. The decision could now be described as independently tested. It gave the executive political cover. The recommendation no longer belonged to one function or one faction. It gave the executive procedural cover. The company could show that a serious process had been followed. It gave the executive reputational cover. The brand on the page carried weight before the content had to stand on its own. It gave the executive blame cover. If the decision failed, management could point to the process, the advisors, the benchmarks, and the approved path. That bundle worked because information was expensive to gather, expensive to structure, and expensive to make presentable. A pyramid of labor could be sold as seriousness. The hours became evidence that work had been done. The volume of the deck became evidence that the problem had been studied. The number of interviews became evidence that the organization had been heard. The premium could hide inside the machinery. AI breaks that hiding place. When research becomes faster, research cannot carry the same price. When synthesis becomes faster, synthesis cannot carry the same price. When slide production becomes faster, slide production cannot carry the same price. When first-pass analysis becomes available to more people inside the enterprise, the outside firm has to prove what remains scarce. The scarce thing is not a chart. It is judgment. The scarce thing is not a benchmark. It is consequence. The scarce thing is not an opinion with a famous logo on top. It is a decision architecture strong enough to turn evidence into action without losing accountability. That is the center of the matter. The old bundle allowed executives to buy help and cover in the same transaction. AI is forcing the bill to be itemized.

Satisficing was not the sin

Herbert Simon taught us something every operator already knows. People do not decide with perfect information. They decide with limited time, limited attention, limited memory, limited Michael Carroll | The One-Degree Dispatch

models, and limited ability to compare every alternative. They search until they find an answer that is good enough to act. They satisfice. That was not an insult to human beings. It was a sober account of how decisions happen in the real world. In a plant, a distribution center, a planning office, a school, a hospital, or a finance meeting, perfect information is not available when the decision is due. The order has to ship. The line has to restart. The supplier has to be chosen. The teacher has to intervene. The cost has to be booked. The truck is waiting. The customer is already angry. Good enough is often the price of movement. No enterprise can optimize every decision. No executive can search every path. No organization can wait until uncertainty is fully removed. Satisficing is not the sin. Institutionalizing it is. The frontline supervisor satisfices because the line is down and people are standing beside a stopped asset. The planner satisfices because the system record is wrong and the real world is already in motion. The plant manager satisfices because the operating plan did not survive contact with the constraint. Those are human responses to pressure. They are visible. They are local. They are often corrected by experience because the consequence appears quickly. Executive satisficing can become different. It can hide inside process. It can select the first acceptable story rather than the truest one. It can choose the first politically survivable option rather than the most durable one. It can accept the recommendation that can get through the board rather than the answer that would force the enterprise to redesign the system. It can call a path responsible because it is defensible, not because it is right. This is where consulting cover and satisficing meet. A consulting engagement can become the stopping rule. Once the firm has studied the issue, once the interviews have been conducted, once the benchmark has been included, once the partner has presented the recommendation, the search is allowed to end. The answer is good enough because the process says it is good enough. The executive is protected because the recommendation came through the accepted channel. Sometimes that is exactly what the enterprise needs. Sometimes the consultant has sharpened the truth, tested it, corrected it, and made it stronger. But sometimes the engagement has only certified the first acceptable answer. It has not deepened the enterprise’s understanding. It has not altered the decision architecture. It has not helped the organization learn how to decide next time. It has only made a good-enough answer safe enough for the executive to carry into the room. Then the room approves it. The enterprise operates it. The bill comes due somewhere else.

Michael Carroll | The One-Degree Dispatch

Where good enough becomes a tax

The cost of executive satisficing rarely appears where the decision was made. It appears downstream. The strategy is good enough for the board, but the operating model cannot carry it. The budget is good enough for finance, but the plant has to reconcile the constraint. The system implementation is good enough for the steering committee, but planners still live in spreadsheets after midnight. The cost program is good enough for the quarterly call, but maintenance inherits the risk. The service promise is good enough for the commercial review, but the supply chain pays with exceptions, expedites, and apology work. This is the COO’s burden. The COO often lives downstream of executive satisficing. The organization approved something that looked acceptable in the room, then asked the system to carry a decision that was not strong enough to operate.

Supporting image. Good enough can survive a meeting and still fail the operating system. The COO often inherits the cost of that mismatch.

That is where the enterprise pays. Not only through consulting fees. Through buffers. Through rework. Through firefighting. Through meetings that exist only because earlier decisions were not clear. Through manual reconciliation between systems that were supposed to connect. Through exception management that becomes normal work. Through human intermediation at the seams. Through the slow erosion of trust when people realize that leadership has approved another promise the architecture cannot keep. Good enough used to survive longer because the environment moved slowly enough to absorb it. A weak decision had time to be corrected by heroics. A delayed decision had time to Michael Carroll | The One-Degree Dispatch

be hidden by inventory. A misaligned system had time to be repaired by people doing invisible work. A poor handoff had time to be recovered before the customer saw the damage. That world is smaller now. Markets move faster than annual planning. Customer expectations move faster than governance. Supply chains break faster than escalation paths. AI changes the work faster than most leadership teams can decide what questions they should now be asking. The carrying cost of good enough has risen. This is why the old consulting model is under pressure. AI does not simply make advice cheaper. It makes weak decision architecture more visible. If research is cheap, lack of information is a weaker excuse. If synthesis is cheap, lack of a storyline is a weaker excuse. If scenario generation is cheap, lack of options is a weaker excuse. If slide production is cheap, the deck is no longer proof of seriousness. The question changes. Why are we still unable to decide? That question points past consulting. It points into the enterprise itself.

Outcome pricing will not save us if causality is missing

The market’s first answer is predictable. If effort is no longer the right proxy, move to outcomes. Pay for cost reduction. Pay for revenue growth. Pay for margin expansion. Pay for cycle time reduction. Pay for business results instead of hours. That sounds better because it is better than paying blindly for effort. But it is not enough. An outcome without a causal chain can become another form of cover. The number moves, but the enterprise may not know why. The firm claims value, the executive claims success, the board sees the metric, and the system absorbs whatever damage was required to move it. A cost target can be hit by starving maintenance. Working capital can improve while suppliers weaken. Productivity can rise while people burn out. Revenue can grow by buying bad margin. Service can improve on paper while exceptions are hidden in manual work. Headcount can fall while contractors, overtime, rework, and escalation labor replace the visible savings. A number can move while the enterprise becomes weaker. That is not value creation. It is metric movement without moral accounting. Outcome pricing only becomes better when the cause of the outcome can be traced. What did we believe would happen? What intervention did we make? What changed? What else could have caused the change? What second-order cost appeared? What capability remained? What did the organization learn? What decision will be easier next time because of the work?

Michael Carroll | The One-Degree Dispatch

Without those questions, outcome pricing becomes a cleaner costume for the same old transaction. The enterprise still pays. The executive still receives cover. The advisor still points to a result. The deeper system remains unexamined. That is why data is not evidence until there is a question. A metric is not proof until the causal chain has been tested. A recommendation is not accountability. A moved number is not transformation. An invoice tied to an outcome is not automatically more honest than an invoice tied to hours. The future of consulting cannot be only outcome pricing. It has to be causal accountability.

The answer is not less consulting. It is Critical Decision Infrastructure

The enterprise does not need fewer advisors as a matter of ideology. It needs better reasons to hire them and better tests for what remains after they leave.

Supporting image. The central question is not whether the deck was polished. It is whether the enterprise became more capable after paying the bill.

The answer is Critical Decision Infrastructure.

Michael Carroll | The One-Degree Dispatch

Figure 2. Critical Decision Infrastructure. The control layer between knowing and acting in an AI-enabled enterprise.

Critical Decision Infrastructure is the operating layer that allows an enterprise to turn evidence into legitimate action. It is not another committee. It is not another dashboard. It is not a transformation office with better language. It is not a governance ritual with AI inserted into it. It is the infrastructure that defines how high-consequence decisions become knowable, permissible, traceable, and owned. It begins with the question. What are we trying to know? What are we trying to make true? What outcome are we trying to shape? What evidence would change our mind? What causal assumption are we making? What intervention should change the condition? What authority is required to act? Where may AI assist? Where may AI recommend? Where may AI execute? Where must AI stop? Where must a human carry judgment by name? How will the decision be recorded? How will the outcome be audited? How will the system learn if it was wrong? That is the missing layer between insight and action. For years, companies invested in data infrastructure. Then analytics. Then dashboards. Then SaaS. Then transformation offices. Then AI pilots. Now they are investing in agents. But many still lack the infrastructure that determines whether any of those assets can legitimately shape an outcome. The system can see more than it can decide. It can report more than it can change. It can escalate more than it can resolve. It can automate tasks without clarifying authority. It can generate recommendations without producing permission. It can create speed without creating legitimacy.

Michael Carroll | The One-Degree Dispatch

Critical Decision Infrastructure closes that gap. It is the architecture of permission made operational. It gives the enterprise a way to move from signal to evidence, from evidence to hypothesis, from hypothesis to authorized intervention, from intervention to accountable action, from action to audit, and from audit to learning. That is the alternative to borrowed courage. A famous consulting logo says, they recommended it. Critical Decision Infrastructure says, we know why we are doing it. A deck says, the process was followed. Critical Decision Infrastructure says, the evidence met the threshold. A board approval says, permission was granted. Critical Decision Infrastructure says, the decision right was clear, the assumptions were visible, the intervention was traceable, and the consequence was owned. That is a different kind of legitimacy. It cannot be purchased at the end of a project. It has to be built into the way the enterprise decides.

AI makes this unavoidable

Agentic AI raises the stakes because software is beginning to participate in the shaping of outcomes. An agent must be able to shape an outcome, otherwise it is not an agent. But once software can shape an outcome, the enterprise has a duty to govern the conditions under which that outcome is shaped. This is where the old consulting problem and the AI problem become the same problem. Both can be used to create distance from consequence. The old phrase was, the consultants said. The new phrase may become, the model said. Both can be used to avoid judgment. Both can turn decision-making into theater if the human being with authority steps back from ownership.

Michael Carroll | The One-Degree Dispatch

Supporting image. The phrase the model said must not become the new version of the consultants said.

A model can summarize evidence. It cannot decide what should matter. A model can recommend action. It cannot carry the moral weight of the action. A model can detect a pattern. It cannot stand before the people affected by the intervention and own what happens next. A model can reduce work. It cannot absorb responsibility. This is why the future must not be human out of the loop, and it must not be human as rubber stamp. The right architecture is judgment in the loop. Real judgment. Named judgment. Accountable judgment. The machine can help carry the work. People must still carry responsibility. That distinction is not sentimental. It is structural. If AI systems are going to support enterprise decisions, they need permission architecture, trust architecture, auditability, causal records, and clear boundaries. They need to preserve the chain between question, evidence, intervention, outcome, and accountability. Without that chain, AI becomes the next consulting firm. Not in function. In excuse. McKinsey’s own writing on agentic AI at scale points in this direction by emphasizing data architecture, governance, access control, lineage, interoperability, and human supervision as agentic systems connect models, tools, and workflows. That is not a technology footnote. It is the beginning of institutional legitimacy in a world where software may act before a human committee can meet. But governance alone is not enough if it governs the wrong thing. The enterprise must govern decisions, not only models. It must govern permission, not only access. It must govern evidence, not only data. It must govern consequence, not only output. That is Critical Decision Infrastructure.

What consulting has to become

The consulting firm that wants to matter in this world should not defend the old bundle. It should take the bundle apart and price it honestly. Research will be cheaper. Synthesis will be cheaper. Drafting will be cheaper. Presentation will be cheaper. Generic strategy language will be cheaper. Benchmarking will be easier for clients to challenge. Anything whose price depended mainly on time, scarcity of access, or the cost of producing the artifact will face pressure. But judgment should not be cheaper. Causal reasoning should not be cheaper. Operating experience should not be cheaper. Trust should not be cheaper. Implementation capability should not be cheaper. The ability to build durable decision infrastructure should not be cheaper because most enterprises do not have it. Michael Carroll | The One-Degree Dispatch

The firms that understand this will not sell faster decks. They will help companies build the machinery that makes better decisions possible after the consultants leave. They will show the causal chain. They will show the decision rights clarified. They will show the evidence thresholds built. They will show the AI permissions governed. They will show the audit trail created. They will show the capability transferred. They will show the cycle time removed from decisions without removing accountability. They will show how the enterprise can now do something it could not do before. If they cannot show that, the price should fall.

Figure 3. When Good Enough Becomes Expensive. The enterprise pays when acceptable decisions are approved before the operating system can carry them.

That is not anti-consulting. It is pro-value. A serious advisor should welcome this standard because it separates the firms that create capability from the firms that sell institutional protection. It separates expertise from theater. It separates implementation from ceremony. It separates decision infrastructure from deck production. The client has to change too. Boards and executive teams should stop asking only which firm to hire. They should ask what decision they are unable to make without the firm and why. If the answer is expertise, buy expertise. If the answer is capacity, buy capacity. If the answer is external challenge, buy challenge. If the answer is implementation help, buy implementation help. If the answer is causal discipline, put it in the scope. If the answer is Critical Decision Infrastructure, make that the work. Michael Carroll | The One-Degree Dispatch

But if the answer is political courage, do not hide it inside a statement of work. Say it. We do not trust our own evidence. We are not aligned. We know the condition, but we do not have the authority path to act. We have built an enterprise that can generate information but cannot turn information into authorized consequence. That honesty would not weaken the enterprise. It would strengthen it.

The executive cannot outsource consequence

A leader can hire advisors. A leader can buy analysis. A leader can commission benchmarks. A leader can seek outside perspective. A leader can ask to be challenged. A leader can admit uncertainty. All of that is legitimate.

Advice can be bought. Accountability cannot

But a leader cannot outsource consequence. That is the line the old model blurred. The consulting firm can recommend a plant closure, but the executive has to carry the human and operational consequence. The consulting firm can recommend a cost program, but the executive has to live with the maintenance risk, the morale damage, and the customer effect. The consulting firm can recommend a new operating model, but the executive has to decide whether the structure improves the flow of work or merely moves boxes on a page. The consulting firm can recommend a strategy, but the executive has to own the tradeoff. The best executives already know this. They use consultants as instruments, not shields. They take the outside view seriously, but they do not hide behind it. They understand that advice is input, not ownership. They understand that a recommendation is not a decision until a human being with authority accepts the burden of acting. The weaker executives use consulting differently. They use it to create distance between themselves and consequence. They use it to make the decision look inevitable. They use it to carry politics. They use it to tell the board that the hard move was not merely their judgment, but the conclusion of a process. This is why the title matters.

Michael Carroll | The One-Degree Dispatch

The cover was never for the enterprise. The enterprise did not need protection from its own truth. The executives did. And the enterprise paid. AI is making that harder to hide. It can write the deck. It can summarize the evidence. It can produce options. It can generate scenarios. It can draft the recommendation. It can make the artifact cheaper and faster. But it cannot stand in the room and say, this is our decision, this is what we believe, this is what we will do, this is how we will know if we are wrong, and this is who owns the next action. That remains human work.

A prediction that can be tested

Here is the testable prediction. Within the next thirty-six months, the strongest advisory contracts will begin to look less like traditional consulting scopes and more like decision infrastructure agreements. They will still include expertise, analysis, and implementation support. But they will also specify the decision rights to be clarified, the evidence thresholds to be built, the causal assumptions to be tested, the AI permissions to be governed, the audit trails to be created, and the capability that must remain after the work is done. If that does not happen, clients may still win lower rates. They may not get better decisions. They will win the rate card and lose the enterprise. The firms that understand this will not defend the old pyramid. They will move toward the work that still matters when the artifact is cheap. They will help clients build the machinery of legitimate action. The firms that miss it will keep selling faster versions of the same thing. Cheaper research. Faster slides. More polished recommendations. Less defensible value. That market will not disappear overnight. Prestige dies slowly because fear keeps it alive. Boards will still want cover. Executives will still want protection. Enterprises will still confuse a famous logo with a safer decision. But the pressure will build because AI has changed the economics of the artifact. Now the economics of courage will be forced into view.

The room after the room

Return to the boardroom after everyone has left. The chairs are still slightly turned from the meeting. The screen is dark. Someone has left a marked-up packet on the table. The recommendation has been accepted. The next steps have

Michael Carroll | The One-Degree Dispatch

been assigned. The invoice will move through accounts payable. The enterprise will begin to carry what the room decided. That is where the truth of the work will appear. Not in the elegance of the deck. Not in the confidence of the partner. Not in the comfort of the board minutes. Not in the claim that the process was followed. The truth will appear in whether the operating system can carry the decision without converting it into more delay, more reconciliation, more exception work, and more hidden cost. For a long time, the invoice was bundled. Advice. Strategy. Transformation. Benchmarking. Implementation support. Program management. Board materials. Executive alignment. Buried inside that invoice was a quieter line item. Cover. AI is forcing that line item into view. Research is cheaper. Slides are cheaper. Summaries are cheaper. Generic analysis is cheaper. Benchmarking is easier to challenge. Information is cheaper. But judgment is not cheaper. Causality is not cheaper. Trust is not cheaper. Execution is not cheaper. Human accountability is not cheaper. Legitimate authority is not cheaper. Critical Decision Infrastructure is not cheaper because most enterprises have not built it yet. AI will not eliminate the need for advice. It will eliminate the ability to hide weak value inside expensive production. It will not eliminate consulting. It will separate consulting that creates capability from consulting that sells institutional protection. It will not eliminate executives. It will expose which executives are willing to own consequence and which ones were renting courage with enterprise money. That may be painful. It should be. The boardroom silence will not disappear. There will still be moments when everyone knows the answer and no one wants to own it. There will still be leaders who want a famous name on the page before they act. There will still be boards that prefer a process they can defend over a truth they must face. But the excuse is getting thinner. The machine can write the deck. The machine can summarize the evidence. The machine can model the scenarios. The machine can recommend a path. What it cannot do is carry the burden of consequence. AI is not merely repricing consulting. It is repricing avoidance. It is exposing the cost of executive cover. It is exposing the limits of satisficing when good enough becomes a tax on the enterprise. It is forcing companies to build Critical Decision Infrastructure so judgment can move at the speed of consequence without losing legitimacy.

Michael Carroll | The One-Degree Dispatch

AI is not merely repricing consulting. It is repricing avoidance

The cover was executive. The bill was enterprise. Now the question returns to the people in the room. What do we know? What must we make true? What are we willing to own? And what infrastructure must we build so we never again have to buy cover for decisions we should have been able to make ourselves?

References and intellectual foundation

This article draws on the Financial Times reporting on AI’s pressure on consulting pricing and outcome-based fees; Mark Minevich’s public framing of consulting’s premium as executive cover; McKinsey’s published materials on Lilli, generative AI, and agentic AI foundations; Business Insider reporting on McKinsey’s increasing use of outcome-based pricing; Herbert Simon’s Nobel Prize lecture on bounded rationality, selective search, and satisficing; and Michael Carroll’s continuing work on decision latency, agency, causal reasoning, Critical Decision Infrastructure, the architecture of permission, and the principle that data does not become evidence until there is a question.

Michael Carroll | The One-Degree Dispatch