Framework·Future of Work

Why Repricing Knowledge Work Is Harder Than It Should Be

March 6, 2026·11 min read

Most of what we call knowledge work is formulation. Not producing things… formulating them. Taking messy reality and converting it into structured action. The lawyer doesn’t produce documents. The strategist doesn’t make decks. The PM doesn’t create status reports. They formulate: reading a situation, applying judgment, and building something that commits real people to real consequences. Jack Skeels wrote a sharp piece recently called “What AI Actually Breaks When You Sell Thinking for a Living” that names what AI is actually disrupting here. Not the artifacts. The formulation underneath them. And he draws a distinction I think anyone in professional services needs to sit with. He calls it the assembler-architect split: Assemblers operate on known patterns. They recombine existing frameworks, structures, and rhetorical moves to fit recognizable situations. Skilled work. Valuable work. But the materials already exist in recognizable form. Architects operate on real conditions. They read the actual situation, including the politics nobody wrote down, the unstated constraints, what’s already been tried and why it failed, and build something that fits that, not a template. AI can assemble. It can’t reliably architect. And for a long time, both were bundled together and priced as one thing. AI is unbundling them. Skeels is right about this, and his full framework (including a four-quadrant model and the Architect-Assembler-Architect loop) is worth reading in full. I’m not going to rehash it here. I’ll use his language loosely, though on my side of the fence the “assembler” work looks more like execution: the production, the deck builds, the deliverable grind… and the “architect” is just the person in the room doing the judgment work that nobody itemizes. What I want to explore is a dimension Skeels doesn’t cover: the reason the repricing is so hard isn’t just commercial. It’s psychological. And understanding why requires looking at professional identity and incentive design, not just pricing strategy.

The Pricing Conundrum Everyone Sees

Skeels nails this part. In an hours-based model, the architect’s judgment is structurally invisible. When you fix a bad brief, catch three contradictions in a strategy, or restructure an entire approach before execution begins… that’s the most valuable work in the engagement. But in a model where revenue equals hours times rate, doing that work quickly and well actually shrinks your revenue line. So practitioners learn to do it quietly. They fix the brief, smooth the incoherence, absorb the contradictions, and never name it. The judgment gets bundled into the price of execution. Clients never pay for it explicitly. They get it “for free”. When AI makes execution cheap, that bundled judgment disappears with it. And now agencies (and law firms, and consultancies, and every other formulation business) face what Skeels calls the repricing problem: convincing clients to pay for something those clients never knew they were receiving. The pricing model has to change before the judgment becomes safe to surface. That’s the conundrum most people are focused on right now, and rightly so. But there’s another one tangled up in it. And it’s the one I keep running into.

The Identity Conundrum Underneath It

Say you change the pricing model tomorrow: Deliverables-based pricing, outcomes-based pricing. You’ve solved the commercial problem. But wait, there’s more. The person who spent 15 years building their professional worth around the execution… the deep research, the meticulous deck builds, the 80-hour weeks, the craft of grinding through complex analysis and more deck builds… that person still thinks of themselves as the one who does the hard, thorough work. That’s not just what they do. It’s who they are. And the new model is asking them to let go of the very thing they’ve been valued (and have valued themselves) for. Letting go of that isn’t a skill problem. It’s closer to grief. This isn’t stubbornness. It isn’t resistance to change. It’s a predictable psychological response that researchers have documented across completely different domains. Chen & Reay (2021) studied professionals facing imposed role changes and found four stages that unfold in a specific order:

Resisting and mourning the loss of previous work. Not a skill gap. Grief.
Conserving professional identity and avoiding the new work. Self-preservation, not stubbornness.
Parking professional identity and learning. Only after grief is processed can genuine learning begin.
Retrieving and modifying professional identity. People don’t adopt entirely new identities. They reach back and reshape the one they had. Stages 1 and 2 are not about competence. No tool training addresses them. No L&D program reaches them. People have to mourn before they can learn. William Bridges’ Transition Model makes the same point from a different angle. He distinguishes between change (the external event) and transition (the internal psychological process). Organizations almost always try to skip from the announcement straight to the new beginning. They roll out the new pricing model, launch the training program, celebrate early wins. They completely bypass what Bridges called the Neutral Zone… the disorienting in-between where the old identity is gone but the new one hasn’t solidified. That’s where the actual psychological work happens. And it looks like low productivity from the outside, which is why leaders rarely protect it. 2040 Digital’s research (2025) brought this into sharp focus with a pharmaceutical company’s AI transformation. Senior scientists who had spent 20 years building expertise in data interpretation became the biggest obstacles to an AI-powered clinical trials initiative. They weren’t opposing the technology. Their identity was built on the expertise AI was compressing. The researchers’ conclusion: traditional change management makes identity disruption worse by celebrating the new system’s benefits (implicitly devaluing old expertise), treating hesitation as resistance, and providing more training when what’s needed is identity support. The pattern keeps showing up. Veterans returning from combat. Retirees losing their professional role. Professionals facing imposed job redesign. The mechanism is the same: when the external scaffolding of identity gets disrupted, people don’t need more training. They need space to grieve, permission to be temporarily incompetent, and help reconnecting to why they do the work… not just how.

The Mechanism: Why the Pricing Model IS the Identity Intervention

Here’s where it gets interesting. And where I think these two conundrums aren’t actually separate problems. Behavioral economist Uri Gneezy’s research (Mixed Signals, 2023) reveals the causal mechanism: when words and incentives conflict, people follow the incentives. Every time. And incentives don’t just shape behavior. They shape how people see themselves. Consider what happens when an architect uses AI to reshape what used to be a 30-hour engagement in an afternoon: In an hours model, the self-signal is: I just made myself less valuable. The rational response is to hide the efficiency. Don’t tell anyone you reshaped the brief in a few hours. Stretch the work to fill the week. The judgment stays invisible because making it visible is career-threatening. In a deliverable model, the self-signal shifts: My judgment enabled this. The efficiency becomes a feature, not a bug. You delivered the thing. How you got there is your craft, your advantage. The judgment starts to become nameable because the model rewards the output, not the hours. In an outcome model, the self-signal shifts again: I freed the rest of my week for the strategic thinking where my real value lives. The judgment isn’t just safe to name, it’s what you’re being measured on. The model actively encourages you to spend time on the highest-judgment work because that’s what drives the outcomes you’re being compensated for. Same person. Same AI tool. Same work. Three completely different identity experiences determined entirely by the business model surrounding the behavior. The pricing model isn’t parallel to the identity shift. It IS the identity intervention. They aren’t two separate problems that need to be “aligned.” They’re dancing together whether you choreograph them or not. Best not to trip! Gneezy’s most famous study makes the danger vivid: a daycare introduced a fine for late pickups. Lateness doubled. The fine converted a social obligation (“I should respect the teachers’ time”) into a cheap transaction (“I can buy extra minutes”). When the fine was removed, the parents stayed late. The social norm didn’t return. The lesson for transformation leaders: the first incentive you introduce doesn’t just change behavior. It changes how people interpret the behavior. Get it wrong, and you may not be able to get the original motivation back. Some doors only open one way.

Why They Have to Move Together

All vision, no path. That’s what most transformation programs feel like from the inside. Change the pricing model without addressing identity, and you get surface compliance. People bill differently but still define themselves by execution volume. They hit the new metrics while privately mourning the old ones. The transformation looks successful on dashboards and feels hollow in the hallways. Address identity without changing the pricing model, and you get frustrated architects who see the shift clearly but can’t act on it without shrinking their own revenue. They know their judgment is the valuable part. The system still pays them by the hour. Eventually they leave, and the judgment walks out with them. Address neither and just train people on AI tools, and you get what most organizations have right now: adoption metrics that look healthy (“90% weekly AI usage!”) sitting alongside stalled transformation and quiet disengagement from your most experienced people. The ones with the deepest judgment are often the ones with the most identity at stake. The uncomfortable truth: the people most likely to resist aren’t your least capable. They’re often your most experienced, most invested practitioners… the ones who built the deepest sense of professional worth around the craft that’s being disrupted. Losing them isn’t just attrition. It’s losing the judgment the new model depends on.

What This Looks Like in Practice

I’m not going to pretend I have this figured out. I’m just a guy with a lot of curiosity who sees a lot of connections. These are patterns I’m seeing and testing, not proven solutions. But for what they’re worth: Name the loss before you sell the future. People need to hear something specific before they can engage with transformation: “The craft you built is real. The years weren’t wasted. What’s changing is how your judgment gets deployed, not whether it matters.” This isn’t a line in a town hall deck. It’s an ongoing conversation that has to happen at the team level, with enough specificity that people feel their particular expertise acknowledged, not just “everyone’s contributions.” Watch your first consequences. The organization’s response to the first person who uses AI to do exceptional work in half the time and the first person whose AI experiment fails… sets the cultural trajectory for the entire transformation. One data point. Cascading consequences. If the fast architect gets their hours questioned (“how can it be good in a few hours when it used to take 30?”), every other architect gets the message. If the failed experiment gets punished rather than studied, every future experiment goes underground. Stop measuring adoption. Start measuring what adoption is supposed to produce. “90% weekly AI usage” as a transformation KPI is measuring the means, not the end. It tells you people are opening the tools. It tells you nothing about whether the tools are enabling better judgment, better outcomes, or better work. When the means becomes the metric, you get compliance theater. Protect the social norms you already have. Craft pride, mentoring culture, quality obsession… these were built over years of execution-era practice. They’re valuable. And they can be destroyed in a quarter by a badly designed incentive that signals “speed over craft” or “volume over judgment.” Audit what’s working before introducing new metrics. Give people agency in how they engage. Research on forced vs. voluntary retirement shows significantly more identity distress when the change feels imposed rather than chosen. The same applies here. “You need to change or you’ll be left behind” produces more resistance than “you’re already doing the judgment work… let’s make it visible and valued.” Frame evolution, not revolution.

The Question Underneath the Question

Skeels ends his piece with a sharp question: Are you assembling, or are you architecting? It’s the right question. But I think there’s a harder one underneath it: What happens when someone answers honestly and the system around them isn’t ready to hear it? The architect who says “I reshaped your brief in an afternoon because I’ve spent 15 years learning to see what’s broken” is telling the truth. But in a system that prices hours, that truth is dangerous. And in a professional culture that equates effort with value, that truth threatens everyone around them too. The pricing and the identity have to move together. The business model has to make judgment commercially safe to surface. And the person doing the judging has to feel safe letting go of the execution grind as the thing that defines their professional worth. Neither one moves easily. Both require patience, specificity, and a willingness to sit in the uncomfortable middle where the old model is clearly broken but the new one hasn’t solidified yet. Bridges called that the Neutral Zone. I’d call it the place where the real transformation happens, or as my son’s Middle School Open House called it, “productive discomfort”.

Sources referenced:

Skeels, J. (2026). “What AI Actually Breaks When You Sell Thinking for a Living.”
Chen, Y. & Reay, T. (2021). “Responding to imposed job redesign.” Human Relations, 74(10).
Bridges, W. (1991/2017). Managing Transitions.
Gneezy, U. (2023). Mixed Signals: How Incentives Really Work.
Novak, K. (2025). “The Identity Crisis of Expertise.” 2040 Digital.
Fadeeva, A. et al. (2025). “Retirement Adjustment Framework.”