The proposal lands in your inbox on a Tuesday morning. You open it, scroll past the cover page, and go straight to the fees section. The numbers are familiar. Not identical to last year, perhaps a modest uplift for inflation, but the same shape. Same day rates. Same timeline. Same structure of how many people doing what for how long.
You have worked with these people before. They know your business. The work needs doing.
So you approve it, forward it to procurement, and get on with your day.
But there is something you cannot quite name sitting at the edge of that decision. Not suspicion exactly. More like a question you have not yet found the words for.
The thing we stopped questioning
For as long as professional services have existed, the invoice has been built on a single unit of exchange: human time.
Not outcomes. Not intelligence. Not results. Time.
The hourly rate. The day rate. The retainer. The project fee calculated by estimating how many people will sit with your problem for how many days. We accepted this model so completely that we stopped seeing it as a model at all. It became the natural order of things. You need something done. Someone does it. You pay for the hours it took.
The reason this model took hold is simple and entirely logical. Time was the only thing you could measure reliably. You could not invoice for thinking. You could not put a meter on judgement. You could not bill for the moment a senior strategist connected two things in their mind that no one else had connected. So the industry settled on time as a proxy for all of that. It was imperfect. Everyone knew it was imperfect. But it worked well enough for long enough that the imperfection became invisible.
Here is the reframe.
You were never actually paying for time. You were paying for the cognitive labour required to produce an outcome. Time was just the measuring stick. The hours were a proxy for the expertise, the knowledge, the accumulated experience that lived inside the people doing the work. When a senior lawyer spent three days on your contract, you were not paying for seventy-two hours of their existence. You were paying for twenty years of pattern recognition, compressed into a document that protected your interests.
That distinction mattered less when humans were the only ones doing the thinking.
It matters enormously now.
What just changed
AI has not simply made tasks faster. That framing undersells what has actually happened.
What AI has done is decouple expertise from time. For the first time in the history of knowledge work, the cognitive labour that used to require a trained human sitting with a problem for hours or days can now be initiated, accelerated, and in many cases largely completed in a fraction of that time. The expertise is still required to direct it, to quality control it, to apply it to the specific texture of a real situation. But the time is gone.
Think about what that means across the professions.
A lawyer drafting a commercial agreement used to spend two days on the first version. Not because they were slow, but because reading, synthesising, structuring, and drafting a complex document to a professional standard takes that long when a human does every step. That same lawyer, working with AI tools that understand legal language at a sophisticated level, can now produce a first version in forty minutes and spend the remainder of their time on the genuinely difficult parts: the edge cases, the client-specific risk, the negotiating position.
A developer building a standard web application used to estimate eight to twelve weeks. That estimate was honest. It reflected how long the work actually took. The same application, built with AI-assisted code generation, can be scaffolded and iterated in a fraction of that time. The developer’s expertise is still there. Their judgement about architecture, security, and user experience is still essential. But the volume of time required to translate that expertise into working software has compressed dramatically.
A strategist producing a market analysis used to need two weeks to research, synthesise, and present a defensible view. The same strategist, using AI to accelerate the research and synthesis, can produce the same quality of output in three days and use the time saved to test more scenarios, explore more angles, or deliver more depth than the original brief asked for.
This is happening across legal, creative, financial, technical, strategic, and operational work simultaneously. Not in the future. Now.
The uncomfortable arithmetic
So here is the question that the invoice sitting on your desk is asking you, whether it knows it or not.
If the work that used to take ten days now takes three, what are you paying for?
There are really only three honest answers.
The first is that the supplier has adopted AI and is passing the benefit to you in some form: lower cost, faster delivery, greater depth, higher quality, more iterations, better outcomes. The commercial relationship has evolved to reflect the new reality of what is possible.
The second is that the supplier has adopted AI and is not passing the benefit to you. The work takes three days. The invoice says ten. The margin has expanded significantly. You are paying the old price for the new speed, and no one has mentioned it. This is the most common situation right now. Not because suppliers are dishonest, but because no one has yet been asked the question directly and the existing model is too comfortable to disrupt voluntarily.
The third is that the supplier has not adopted AI at all. The work still takes ten days because it is still being done the old way. In that case you are paying the right price for the wrong era. And the gap between what you are receiving and what is now possible is growing every month.
None of this is a reason to distrust your suppliers. Most of them are navigating genuinely complex transitions. AI adoption inside a professional services firm is not as simple as installing a tool. It involves retraining, restructuring workflows, managing quality control in new ways, and in many cases rethinking the fundamental shape of the service being delivered. That takes time and deserves acknowledgement.
But the conversation still needs to happen.
What clients need to understand
If you are buying professional services of any kind, the most important shift in your thinking is this: stop buying time and start buying outcomes.
This is easier said than done because the entire commercial architecture of most supplier relationships is built around time. Contracts are written in day rates. Proposals are structured around resource allocation. Conversations about scope default to discussions about how many people for how many weeks.
You will need to be the one who changes that conversation, because most suppliers will not initiate it voluntarily. Not out of dishonesty, but because the existing model is comfortable, familiar, and financially convenient for them.
The questions worth asking at every supplier review, every contract renewal, every new proposal are these.
What does AI now enable you to deliver that you could not deliver eighteen months ago? How has your use of AI changed the time it takes to produce this work? If the time has compressed, how is that reflected in what you are charging or what you are delivering? What does the scope look like if we price for outcomes rather than hours?
These are not aggressive questions. They are the questions of an informed buyer operating in a world that has changed. Any supplier who cannot engage with them seriously is telling you something important about where they are in their own evolution.
What suppliers need to understand
If you are delivering professional services of any kind, the most important shift in your thinking is this: the defence of your existing pricing model is not a long-term strategy.
The instinct when this conversation comes up is to reach for the expertise argument. You are not just paying for the task. You are paying for the judgement, the experience, the twenty years of pattern recognition that no model can replicate. That argument is partially true. It is also becoming insufficient on its own.
Because here is what it misses.
AI has not eliminated expertise. It has changed where expertise is applied. If a model can now handle sixty percent of the cognitive labour that used to fill your days, you have been handed something remarkable: the capacity to apply your expertise to more problems, at greater depth, with better testing and iteration, and to more clients simultaneously.
The question is not whether your expertise is still valuable. It is. The question is whether you are monetising it intelligently in a world where the mechanics of delivery have fundamentally changed.
The suppliers who will define the next decade are not the ones defending the old model. They are the ones who have understood that AI has expanded what they can deliver and have restructured their offering around outcomes, quality, and scale rather than time. They are charging for what they can now achieve, not for what it used to cost to achieve it.
That is a different and ultimately more powerful commercial position. It is also an honest one.
Where the value actually lives now
The knowledge compression that AI represents is real and significant. A meaningful portion of what professional services firms have historically charged for, the research, the drafting, the synthesis, the standard analysis, now lives inside models that anyone can access. Not all of it. But enough of it to permanently change the baseline of what clients should expect to receive for their investment.
The value that remains irreplaceable sits in three places.
The first is judgement under genuine ambiguity. AI is exceptional at processing what is known. It is weak at navigating what is genuinely uncertain, politically complex, or dependent on the kind of human reading of a situation that comes from being in the room. The adviser who can sense what is not being said, make a call when the data is incomplete, and take accountability for the outcome still has something that no model can offer. That is a smaller and more specific slice of the work than it used to be. It is also the most valuable slice.
The second is execution at a new altitude. If AI compresses the cognitive labour of delivery, the intelligent supplier is not doing less work. They are delivering more. More scenarios tested. More iterations produced. More depth in the analysis. More versions of the creative work explored before a recommendation is made. More time spent on the hard questions because the routine work has been accelerated. The client who is buying well should be receiving more for the same investment, or the same for a lower investment, depending on the nature of the relationship.
The third is the orchestration of intelligence itself. Someone has to know how to direct the AI, quality control its output, connect it to the specific context of a client’s situation, and take accountability for what gets delivered. That orchestration capability is genuinely scarce and genuinely valuable. It is not something clients can buy off the shelf. It is something suppliers build through practice, through iteration, and through developing real expertise in how to deploy these tools for specific types of problems. That is worth paying for. It is different from what was worth paying for before. The conversation needs to catch up.
The invoice that has not changed
Go back to that proposal sitting on your desk.
The numbers are not wrong because the supplier is dishonest. In most cases they reflect a model that made complete sense until very recently and that no one has yet had the explicit conversation to update.
That conversation is overdue on both sides.
Clients who do not ask the question are leaving value on the table, paying old prices for new capabilities, and missing the opportunity to get dramatically more from the suppliers they already trust.
Suppliers who do not initiate the question are building on a foundation that is less stable than it looks. The clients who understand what AI makes possible are starting to ask. And when they do, the suppliers with no answer will find themselves competing on price alone, which is the worst possible position for a knowledge business to be in.
The shift is not about squeezing. It is about honesty. And it is about both sides having the maturity to look at a changed world and update their commercial relationship to reflect it.
The model that priced intelligence by the hour made sense when humans were the only ones producing it.
That world has ended, somewhere in the last two years, while the invoices stayed the same.
If this resonated, subscribe. The next piece goes further into what the new commercial model for professional services actually looks like and what both sides need to build for the decade ahead.
Craig Hepburn is an AI strategist and builder, Perplexity Fellow, former CDO at Art Basel and UEFA. Works across tech, business and system design shaping how AI operates responsibly.
A very insightful write-up, thank you Craig.
On my site, I showcase how people can use the different LLMs as partners in thinking with the human being the navigator and pilot.
https://lordstretch.substack.com/p/do-you-have-walk-away-power - as one example.
Employees able to 'pilot' the AIs like this unlocks the employee expertise.
The mechanic who KNOWS because they heard the sound while walking up to you, verses the one who can only do what the computer tells him to do....
Which one do YOU want?