
The Case of the Disappearing Margin
AI now does the junior work that big accounting and consulting firms bill by the hour. Who keeps the saving, the firms or their clients?
The Firm in this report is a composite. Big Four for the audit anchor, the partnership scale and the professional-indemnity exposure. Top-three strategy firm for the project pricing and the ratio of juniors to partners. No real firm is described.
Dr Dan Epstein · version 0.17 · 12 October 2026
A Strategy Soup ScenarioWhat is an exercised scenario?
The question
The subject is professional services, the firms that sell thinking: accountants, auditors, lawyers, management consultants. A big one is shaped like a pyramid: a few partners at the top, who own the firm, many junior analysts at the bottom, and profit that comes from billing junior hours at several times what the juniors cost. AI now does much of the junior work, so somebody gets the saving: the firms or their clients. Inside a firm the question sounds like this. When AI makes analysis cheap to produce, do we protect the pyramid that pays us, or undercut our own prices before somebody else does?
The answer. In our story the firms keep the saving at first, and their clients have most of it by 2030. The firms keep it first by hiring fewer graduates and leaving the price where it was. Each job has two parts: routine work, about seventy per cent of it in our story, and the part a named, licensed person must check and sign. The client's own staff learn what the routine part costs to make, do more of it in-house, and stop paying the old fee for it. The signed part stays with the firm, at a price. The insurer that covers a firm when its advice turns out to be wrong charges more to cover AI work that no named person has signed, or refuses to cover it, so the firm keeps paying people to sign. , By 2030 the firm keeps the signed part of each job, and its partners keep their income by employing far fewer people. The fork, the point where this story could still go more than one way, is where the price cut lands, and who is blamed when it does.
The roles. The simulation has five roles: the incumbent firm, a challenger firm, the buyer, the workforce, and the insurers and regulators as a single role. The simulation tells one story, from 2026 to 2030. Nothing was decided by chance, and the story was played once, so this report gives no counts of how often anything happens. The report shows one path and argues for it. The real-world record behind the story stops on 1 July 2026, and every later date is the story.
How to read the markers. A marker like points at a checked fact in the Sources table at the end, with its source, a quote from it, and the date the quote was checked. A marker with a small magenta star beside it points at a fact that is true with a caveat, and the Sources table states the caveat. Where a marker sits on a sentence that mixes fact with story, it backs the pattern only. The invented company and the exact figure are ours. A number with no marker is one we chose, and the seams list the main ones.
The players. Five players carry the story we wrote out of those five roles. The Firm is a composite of a Big Four accounting firm (Deloitte, EY, KPMG, PwC) and a top-three strategy firm (McKinsey, Bain, BCG). From the Big Four it takes the audit anchor, the scale of the partnership and the professional-indemnity exposure that comes with signing audits, valuations and tax positions. From the strategy firms it takes project pricing and the ratio of juniors to partners, the pyramid above. We built it from the real firms that cut graduate hiring hardest, so it copies no one company. Throughout, "the Firm" means only this invented company, and "the big firms" means the real Deloitte, EY, KPMG, PwC and the large strategy firms.
The Challenger is a new firm of two to six senior people with no juniors under them. The Buyer is the chief strategy officer of a large company, who decides what work is bought outside and what is done in-house. The Analysts are the Firm's own junior staff, and they carry the workforce role. The Underwriter is the market that carries the Firm's professional-indemnity risk, the cover that pays the client when the Firm's advice is wrong. It carries the insurers' half of the fifth role. The regulators' half arrives late in the story, as the written rules.
AI-heavy work means work where AI produced most of the output. Unsigned AI work means AI-heavy work that no named professional has checked and put their name to.
From here we write as if you run one of these firms, because that is who the simulation was built for. If you do not, read "you" as "a managing partner". What follows is written from 2030, looking back.
2026 H1
The inoculation theatre
The Firm opened with a loud announcement that changed almost nothing, and an unannounced cut that changed a great deal.
The loud move was an alliance with a leading AI lab and a cloud provider, under a ten-figure "AI investment" headline. , , Most of that number was existing technology spending, relabelled. EY's own launch release describes its US$1.4bn in the past tense, as money already spent. , The Firm built an internal delivery platform and put an "AI capability" line on every proposal. It was inoculation: the Firm turned the threat into a service line it could sell, so AI arrived on the books as revenue, and the price stayed where it was.
In the same two quarters it cut its graduate intake, the number of new graduates it hires each year, by fifteen per cent, and stopped replacing the juniors who left. It kept cutting. For every 100 graduates the Firm hired in 2025, it hired 85 in 2026, 42 by 2028 and 33 by 2030, figures we chose. So its cut was fifteen per cent in 2026 and fifty-eight per cent by 2028. Among the real firms the steepest one-year cut was KPMG's, about a third of its UK intake of graduates and apprentices in 2024. , The other three cut less.
Internally that was "raising the bar". On the books it was the one lever that lifted margin at once and in silence: fewer juniors, the same price to the client, and the saved cost going to the partner draw, the money the owners take home. The platform cut what delivery cost the Firm and did not cut what the client paid.
The lift arrives before any change to pricing. On hourly billing the Firm pays a junior for every hour of the year and bills the client only for the hours that land on a job. Cut the juniors and the whole salary goes, while the platform and the remaining staff absorb the billable hours. Paid hours fall faster than the fee. The gain is a cost cut booked against an unchanged price, and it lasts only until the client learns what the hours now cost to make.
2026 headline (composite, an invented line): "The Firm commits $3bn to AI, says graduate hiring will be 'more selective.'"
2026 H1-H2
The buyer runs a secret test
The Buyer, the Firm's largest client, meanwhile commissioned a normal piece of work from the Firm. Behind a closed door she had two of her own people rebuild sixty to seventy per cent of the same piece of work with off-the-shelf AI tools, then compared the two finished versions. She told almost nobody, because she was gathering intelligence and no purchase hung on it. The sixty to seventy per cent is our figure. In the best real measurement of AI on consulting work, AI made consultants faster on some tasks and worse on others. ,
She came away with two things. One was a document to put in front of her chief financial officer later. The other was a new reference price in her own head: what the routine seventy per cent of the work costs her company to make itself. Once the platform is built and paid for, each extra piece of that work costs her close to nothing. That reference price never appeared on an invoice, and it is the number that later cuts the Firm's fee.
2026 H1
The challenger plants a guaranteed flag
The Challenger was two to six senior operators, no pyramid, no brand. It turned one painfully specific job into a fixed product: commercial due diligence for deals under thirty million dollars, ten business days, thirty-five thousand dollars fixed, money back if it missed. Commercial due diligence is the research a buyer commissions before it buys a company. Deals under thirty million dollars sit at the floor of that market, so the price comparison this report builds on this product is a floor case. Small-deal pricing shows the lowest price the work can carry and says nothing about the fees on large deals.
The guarantee came first, because a firm nobody has heard of cannot otherwise get past a buyer's fear of being blamed. Then the price: twenty-five to forty per cent of what a big firm charges for the same small-deal job. A tenth would have branded the category cheap and left nothing to pay for building a reputation. A third sat in the one band the Firm could not follow, because the Firm still has to cover its juniors, its offices and its partner draw. We chose the price, the band and the guarantee.
Thirty-five thousand dollars at that share implies a big-firm fee of about ninety thousand to a hundred and forty thousand dollars for a small deal. We derived that range and did not check it. The best-documented real challenger charges a twentieth to a tenth of the big firms' fee, measured against their large-deal price of half a million to a million dollars. , The two shares rest on different big-firm fees, and each bar on the chart names its own. So the real challengers price lower and promise less than ours.
2026 H2 - 2027
The insurer moves first, and nobody is watching
The insurer set the first rules for AI-heavy consulting, ahead of any regulator, and it did it by changing a price.
Every professional firm carries professional-indemnity cover, the insurance that pays the client when the firm's advice turns out to be wrong. A small firm buys it from an insurer and gets a renewal quote every twelve months. A firm the size of the Firm funds it through a captive: an insurer the Firm owns, which holds capital against the first slice of every claim, called the retention, and buys reinsurance for losses above that slice. This report assumes the Firm runs a captive. For such a firm the price of risk arrives as capital the captive has to hold and as the price and wording of the reinsurance above the retention, never as a quote. Reinsurers reprice that layer every twelve months, and they can carve a whole category of work out of it on their own, with no statute, scandal or consultation.
Through late 2026 and into 2027, AI-heavy work started drawing new questions from the reinsurers and from the actuaries who set the captive's capital. An argument began about whether "silent AI", a policy that never says whether AI-caused losses are covered, already excluded this work. , In the simulation the Underwriter kept the premium flat where the Firm could name a licensed person who owned each output. It raised the premium on unsigned AI work by fifteen to forty per cent at the 2027 renewal, a band we chose, or refused to cover that work. For the Firm that landed as more capital locked in the captive and a dearer, narrower reinsurance layer: a cost on the balance sheet that never appears on an invoice.
Today's real market runs the other way. Insurers give discounts of 5 to 15 per cent to firms with clean claims records. That is a general discount on the cover, not a price on AI work.
The rise itself is a small sum, because this cover costs between a quarter of one per cent and five per cent of a firm's fee income. What slowed the Firm was the refusal. If the reinsurers carve a kind of work out, every loss on it falls on the captive's own capital. So the chief financial officer, who controls the captive's capital, became the person who slowed the Firm's sale of unsigned AI work. Selling unsigned AI work is now the expensive, personally risky path, and paying a human to check and sign the work is the cheap one.
Reinsurers' loss models, meaning their own estimates of what future claims will cost them, set the working policy on AI in consulting for 2026 through 2028. The people who built those models did not think they were regulating anything.
2026-2027
Hold the price, change the story
The Firm stopped showing the client how many hours the work took. Proposals moved to fixed fees, outcome-based or "value-based" pricing, and managed-service subscriptions, which charge a standing fee for work delivered continuously , . All of it was sold as some version of "aligning our incentives with your outcomes." The billable hour, which means charging the client for time, had started to work against the Firm. Keep selling time and eventually the client asks why the task the AI just collapsed still costs two hundred hours.
Cutting the link between price and input let the Firm keep the difference when AI cut delivery hours by sixty per cent, a figure we chose. The best measurements so far are 25 to 30 per cent. ,
Once price no longer follows input, the client is free to ask what the result is worth to her.
2026-2027
The buyer resets the price list at her strongest moment
The Buyer re-cut her outside contracts away from hours and towards results and access. She stood up a small internal team of four to eight people, light on seniority and heavy on tooling. She killed staff-augmentation spending first, before touching anything the board sees. Staff augmentation means renting consultants by the head to sit inside her own team. To the partners she was blunt: "I'm not paying analyst rates for analyst work anymore."
The window was short, because the firms were frightened and had not yet folded their AI savings back into the old price. Her internal team kept the saved margin, and the knowledge that departing consultants used to carry out of the building in their heads.
2026, ongoing
The juniors hide how fast they work
The partners were reading bad numbers about their own business, because their own staff were hiding the real ones.
The Analysts became the most AI-fluent people in the building, and they hid it. , The slide presentation that used to take twenty hours took four, and they paced the work to look like the old speed. Telling the truth meant either a smaller bonus pool or being the obvious candidate for the next cut. The sharpest built reputations of their own on the side: industry teardowns, public templates, a personal brand that did not need the Firm's logo.
If they declare the speed-up, the gain belongs to the Firm. If they hide it, the gain is theirs. So the workforce absorbed the saving before it reached the Firm's accounts, which is why the measured return on AI lagged the demonstrations. , The partners looked at their own numbers and concluded, reasonably, that AI was not changing their economics yet. That stayed true until a challenger with no hidden slack to protect underpriced them in the open.
2027
The Firm builds its cheap unit and starves it
The Firm built its own unit around AI, set up so that a failure could not hurt the core.
Separate brand, lower price, its own accounts, thinly staffed, and aimed at the work the Firm was about to lose anyway: benchmarking, diligence support, market sizing, routine modelling. , The Firm walled it off and kept it out of the pricing conversation, so it could not set a reference price the partnership would have to live with. The partners reasoned that if the unit won the routine work, the Firm still kept a forty to fifty per cent margin on it, and if a challenger won it, the Firm kept nothing, and the challenger's thirty per cent margin grew as it scaled.
Starved of talent, of freedom to set prices, and of partner status, the unit stayed a toy, and it was not ready when the repricing arrived.
2027
One licensed person carries the blame
Then a case landed on a person, and it changed how every partner thought about risk. An AI-produced output turned out to be wrong: an audit, a valuation, a tax position, a filing, and nobody had actually checked it before it went to the client. The penalty fell on a licensed individual, and the AI, the firm's tool and the firm itself carried none of it. A licence here is the professional body's permission to practise, the thing an accountant, an auditor or a lawyer needs in order to work in the field at all. Losing it, which the professions call being struck off, ends the career. Management consultants need no such licence, so in the story the case lands on the accountant, auditor or lawyer who signed the work. That person was struck off, or sued, or both, and the insurance claim was contested because nobody could show a human had checked the work.
Professional bodies needed one case like it to turn the soft 2026 guidance, that the human remains responsible, into something felt in the gut. "The AI did it" was no defence. Every consequential piece of work needs one named person who can be punished when it is wrong, and that person is the one throat to choke.
2027
The buyer splits her supplier list in two
The Buyer made her suppliers compete directly against each other for the same work, every quarter.
A thin top layer of brand-name firms, kept for board-facing cover and for genuinely new thinking that no template can reproduce. Challengers built around AI underneath for everything operational. Both layers got the same brief, and she compared what came back.
The famous logo still protects the board from criticism, and that use is the last to fall. Bidding the identical brief put a dollar figure on that protection, so she could watch whether it was shrinking fast enough to justify moving a firm down a tier.
2027-2028
Pivot up to accountability, and the middle hollows
"AI gives you the analysis. We give you judgement, the board-defensible decision, regulatory cover, indemnity, and someone accountable when it's wrong." That was the Firm's new pitch, and this time it was honest about what it was defending. It leaned into audit, risk, regulation, restructuring and deals, the work where no model can carry the blame. ,
That defended the top of the pyramid and did nothing for the middle. Salaried managers had risen by running teams of juniors, and half their value had been coordinating people who no longer existed. Those who could direct AI-heavy output survived. Those who had passed work between humans washed out. , , The Firm was slow to react for an ugly reason: those people had no vote in the partnership.
2027-2028
The credential gets questioned at the on-ramp
Graduate hiring fell far enough to see from the street. , The MBA bargain came under open question. An MBA is a two-year business degree, usually paid for with debt, and its promise was an entry-level job at a salary that services a hundred-and-fifty-to-two-hundred-thousand-dollar loan. Both ends of that promise were shrinking. , The sharpest twenty-two-year-olds went around the qualification instead, towards AI skills, a specific industry, or building something of their own.
The path was always a financial bet: take the debt, and let the pyramid salary pay it off. When the salary and the exit job both shrink, the bet stops paying. The Firm had, in effect, chosen a 2035 partner shortage to fund 2027 margin, and partners are paid on this year's profit.
2028
The Firm rebuilds the career ladder around one senior and a machine
By 2028 the margin arithmetic had changed for good, and pricing could no longer hide it. So the Firm rebuilt the career model itself. ,
Fewer but bigger engagements. One senior person overseeing AI-heavy output, where a team of juniors used to sit. A thinner manager tier. An unannounced reset of the up-or-out clock, the rule that says get promoted on time or leave, and a lower rate of admission to the partnership. The people pushed out were the partners' own former protégés.
Removing the salaried middle, whose billable hours no longer covered their cost, protected the partner draw. It happened late, because each cut admitted that the business model had changed.
2028
Politics arrives late and blunt
Years of unannounced hiring cuts turned into a sudden political problem, and nobody had planned for what came out of it. Job losses across the white-collar and graduate professions became a political headline, and politicians reached for blunt instruments: talk of hiring mandates, laws forcing firms to tell clients when AI did the work, training levies. Professional bodies scrambled to sell the qualification as a guarantee of trained human judgement.
The professional bodies also defended the junior analyst pyramid harder than anyone expected, partly for their members and partly for their own fee income, which depends on a steady supply of licensed humans. Fewer juniors trained now means fewer licensed humans later.
2028-2029
The people who left build small firms of their own
Ex-pyramid talent built networked boutiques, meaning small firms that specialise in one thing, and part-time collectives. Pick one industry, turn the work into a fixed product, undercut the Firm on price, match it on quality, on a cost base of three people plus AI, against the Firm's thirty plus AI. , The best moved up-market on purpose and added a thin premium line for a named accountable human.
The binding constraint had moved. It used to be capital, brand and an army of analysts. Once enough trained judgement had left and the tooling had become cheap, the constraint became trust and distribution. Most of these firms still failed the way small firms always fail: they could do the work and they could not be found.
2029-2030
The written rules catch up with the insurer
The binding rules finally arrived in writing, and they surprised nobody who had been watching the insurers. A named accountable human on the record. A duty to tell clients when AI was involved. Audit-trail requirements. A recognised kind of professional-indemnity policy for AI-heavy work, with its own pricing tables. The rules also redefined the qualification itself, from the person who does the work to the person accountable for the work, who certifies the AI did it correctly.
By 2029 the rule-writers had case law, and from the insurance market they had reserves, the money set aside against AI-heavy work, and prices. Settled claims they did not have. A professional-indemnity claim runs years from the wrong advice to the paid loss, so a 2029 loss is invisible in 2029 and the 2027 losses were mostly still open. The rules were written on reserving and pricing, and mostly ratified what the insurers had priced in 2027. The rules permit AI, the firms keep the productivity, and a licensed, insured, disciplinable human stands at the signing point of every consequential output. , The pyramid survived by turning upside down, behind a wall of compliance cost the incumbents could afford and the challengers struggled to clear.
2029-2030
The Firm buys the capability it starved
The Firm acquired the challengers that were winning the fixed-product work, at a price it justified in public and resented in private. The purchase removed a competitor that was setting the price, and bought the capability the Firm had strangled inside its own wall.
The surviving challengers had to choose a lane. Some stayed capped boutiques of fifteen to thirty seniors, took the cash out, and never raised money. Others turned their tooling into software other people could run, a bigger prize and a hard contest against funded competitors and the AI labs. The ones who refused to choose died between the two.
The buyers landed on a standing rule. Do the recurring, embedded work yourself, as the largest companies already do through in-house delivery centres. Save outside spending for the genuinely new, the politically dangerous, and the work that must come from an independent party. The market settled into three tiers: a few expensive accountable judges plus AI at the top, fixed products in the middle , , and the buyers' own internal teams underneath. The old mass-apprenticeship analyst job, the one that made all those partners, is largely gone.
We tracked where the saving ended up on four scales, all of them ours. The fee the client pays per job fell from 100 in 2026 to 50 by 2030. The share of its advisory work the Buyer's company does in-house rose from 4 to 42 per cent. The premium on unsigned AI work rose from 100 in 2026 to 145 by 2030: the first rise was the fifteen to forty per cent band at the 2027 renewal, and later renewals added more. For every 100 graduates the Firm hired in 2025, before its first cut, it hired 33 in 2030. The average partner draw stayed level or rose, because far fewer people now shared the fee.
each crease is pressed harder as the fee the client pays per job falls, on a scale we chose where 2026 is 100: from 100 to 50 by 2030
Where it forks - pick your ending
A handful of hinge points decide which of three endings you reach. Pick the one your own beliefs support, then notice what you had to believe to pick it.
Where it forks - pick your ending
Ending A - The pyramid turns upside down behind a wall of compliance cost (the ending we think most likely)
The Firm does everything in the timeline: the alliance announcement, the intake cuts, fixed fees that keep the cost collapse off the invoice, and a walled-off unit it never lets win. The starting facts are real, the alliance announcements , , the intake cuts and the pricing shift . The motives, and the walled-off unit, are our invention. The insurer and the 2027 liability case make protecting the human-review tier the cheap path, and the Challenger hits a ceiling at the regulated tier. The Firm buys the capability it starved and survives as a smaller, more senior accountability business, earning a higher margin per head. ,
What it looks like: a third of the Firm's 2025 graduate intake, fewer partners, and the average partner draw defended or higher. Revenue lower, gross margin per head higher. The routine seventy per cent of the work goes to the buyers' own staff or to fixed-product firms, one of them the unit the Firm bought. Buyers hire the Firm itself only for the thirty per cent that needs a signature.
The belief it stands on: the AI labs never accept liability for professional-grade output, so a licensed human must. If a lab or an insurer ever credibly covers model output, the signature stops being scarce, the work becomes cheap and interchangeable, and this ending collapses.
Where it forks - pick your ending
Ending B - The buyer does the work itself
A handful of flagship companies very publicly run strategy in-house, with AI and a couple of ex-pyramid hires, and other buyers copy them. Buyers bring in-house anything they use more than twice. The consulting line halves because demand went away, and the challengers won little of it. The Challenger's hardest competitor turns out to be the Buyer's own newly capable team, plus a flood of identical tiny boutiques with no distribution.
What it looks like: company strategy teams double in capability, outside spending falls forty to sixty per cent, and the long tail of staff-augmentation and research vendors disappears. The big firms keep the work that is new, too politically hot to keep in-house, or required to come from outside.
The belief it stands on: buyers build internal capability that lasts, and the hidden fixed costs stay manageable. Platform, data governance, security and the scarce senior people who run it all cost money. Near-zero cost for each extra piece of work has to survive company overhead landing on it.
Where it forks - pick your ending
Ending C - Nothing much changes, and the pyramid pays through 2030
Everyone overestimated how rational buyers are. Procurement inertia, long relationships with the big firms, the preference for a safe pair of hands, and the comfort of a big-firm name on a board presentation keep the money flowing long past the point where the Challenger's economics are obviously better. , , A confidently wrong AI deliverable from a no-name boutique becomes a public disaster and pushes buyer trust backwards. AI stalls on exactly the judgement and taste tasks the pessimists assumed it would take.
What it looks like: the big firms adopt AI to fatten margin while keeping price, and largely get away with it through 2030. Challengers win clients and scrape by, capped by how much senior time they have. The analyst who stayed and served the apprenticeship inherits a thinner but intact firm at higher pay.
The belief it stands on: trust, risk transfer and board psychology are stickier than the economics. Buyers keep paying the brand premium because they were always buying defensibility, someone to blame and political cover. AI made the analysis cheap and left those three alone. If the buyer's reference price resets the moment "AI did the analyst work" becomes consensus, this flips to A or B fast.
Where the real record stops
Everything above is written as memory, and none of it is a record of real events. The real-world facts behind this report stop on 1 July 2026, the "now" the story looks forward from. This version is dated 12 October 2026. Some of the story's late-2026 dates have now passed, and we have not checked the story against what happened after 1 July. The dates are the story's, not a forecast.
When this report was first drafted it named three real signs to watch: insurers writing AI out of their policies at renewal, the first penalty for AI-made errors in consulting landing on a licensed person, the accountant or lawyer who signed the work, and a large company running its strategy work in-house with AI and saying so. By 1 July 2026 only the first had moved: by May 2026 three insurers were seeking approval to exclude AI-driven losses from their policies. The first consulting penalty for AI-made errors had landed on a firm, as a part refund, and on no licensed person. Nobody had yet seen a large company run its strategy work in-house with AI and say so in public.
Choices that look administrative decide which ending a firm walks towards: a hiring number, a change to a pricing model, and whether the AI unit reports to somebody with the power to make it win.
The seams - where this simulation is uncertain
The whole thing flatters human judgement. If AI keeps getting better at judgement, the one thing every role in this report assumes AI cannot do stops being true. Then Endings A and C fall over, because both rest on clients paying a firm for human judgement. All five roles land on the same conclusion, that human judgement, accountability and training juniors under senior supervision are the one advantage a competitor cannot copy. That is suspiciously comfortable, and of the claims the story rests on, it is the one we trust least. The three abilities at stake are framing a problem well, judging what counts as good work, and catching your own confidently wrong output. We will tend to say those belong to humans, because judgement is what we sell. We have left the claim in because we think it is more right than wrong over this window. How much judgement is irreducible we could not settle.
We chose every date after 2027 for the telling. The 2027 liability case, the 2028 political shock, the 2029 written rules: that sequence is clean because we made it clean to tell. Real case law and legislation are lumpy and differ by country. We are fairly confident in the order, insurer first and regulator second. The timing could compress into eighteen months or stretch past 2030, and the claims lag pulls it later, because settled losses arrive years after the reserves and prices the 2029 rule-writers see.
The captive is an assumption. This report assumes the Firm funds its professional-indemnity cover through a captive it owns, with reinsurance above a retention, as large firms commonly do. The pack contains no claim on it. If the Firm bought cover on the open market instead, the same repricing would arrive as a renewal quote, faster and in plainer sight.
The Buyer is the figure most likely to be modelled wrong, in the same direction, by everyone. Every role reasoned as if buyers are rational, price-sensitive and capability-building. Each of the five roles, asked where it was wrong, said it had assumed buyers were more rational than they are. Naming the bias did not remove it. Procurement inertia, board psychology and the value of an outside firm as political cover are the three most likely to be under-weighted across all five roles at once. , A room of real executives already knows how its own board behaves, better than we can guess.
The challenger wave might be a thousand identical boutiques that mostly fail. The roles assume a challenger can build reputation and distribution fast enough to matter, then confess that no distribution is how small firms die. We cannot settle whether the challenger surge is a structural shift or a churn of look-alikes. Do not treat "challengers reset the reference price" as inevitable.
Every number without a marker is one we chose to be consistent with the others, with no measurement behind it. The main ones are the ratios of juniors to partners, the Firm's graduate intake (85, 42 and 33 for every 100 it hired in 2025, a cut of fifteen per cent in 2026, fifty-eight per cent by 2028 and two-thirds by 2030), the sixty-to-seventy per cent the Buyer's team reproduced, the seventy-thirty split between routine work and signed work, the Challenger's thirty-five thousand dollars and its share of the big-firm fee, the fifteen-to-forty per cent insurance repricing, the sixty per cent fall in delivery hours, the four scales in the timeline's last step, the margins on the walled-off unit and the challenger's thirty per cent margin, the two hundred hours of a task the AI collapsed, the slide presentation that took twenty hours and now takes four, the Buyer's internal team of four to eight people, the capped boutiques of fifteen to thirty seniors, and the forty-to-sixty per cent fall in outside spending in Ending B. None is benchmarked to your firm. Replace every invented figure with your own real numbers before you act on any of this.
Where the evidence already cuts against us
In eight places the checked evidence pushes back.
The insurance repricing is a direction of travel. Today's pricing runs the other way. The 2025-26 professional-indemnity market is soft, which means insurers are competing on price. An Australian specialist broker reports discounts of five to fifteen per cent for firms with clean claims histories . Premiums benchmark at roughly a quarter of one per cent up to five per cent of fee income . Our fifteen-to-forty per cent repricing is therefore a multiple applied inside a small band of fee income, and the real money is smaller than the headline reads. The mechanism is loading: by May 2026 three carriers were seeking approval to exclude AI-driven losses from their wordings , and "silent AI" is a named debate in insurance-law commentary , . If the soft cycle persists, the extra cost lands later and lighter than our timeline needs.
The real challengers price lower and promise less than ours. Our Challenger charges thirty-five thousand dollars, a quarter to two-fifths of the fee we assume a big firm charges on a small deal, and matches the big firm's quality. The best-documented live one sells commercial due diligence at about fifty thousand US dollars against a half-million-to-a-million-dollar big-firm job on large deals , and positions itself as "80% of the insight at 5% of the cost" . As a share of the big firms' fee, that is two and a half to eight times cheaper than our band, depending on which ends you compare, and it does not claim to match quality. The two shares are measured against different big-firm fees, and the pack has no checked big-firm fee for a small deal. Nobody in the pack publishes our money-back guarantee , and our story leans hard on it. The closest real analogue is the incumbents' own outcome-linked pricing . The most capitalised real challenger can draw up to three hundred million US dollars of investment from Warburg Pincus, and it aims at large mid-sized companies, well above the cheap end of the market , . Capital has not stopped being a constraint. The pricing move is the weakest-evidenced one in this scenario.
The boutiques may not fail where we say they fail. We have the challenger wave dying of no distribution. Trade coverage of productisation records a different failure mode: independents now lose on whether their delivery is demonstrable, structured and traceable, even when the raw quality is there . The binding constraint may be a provable method plus visibility.
In consulting the blame still lands on the firm. Courts have sanctioned individual licensed attorneys for AI-fabricated citations, including disqualification from a case . But the highest-profile consulting incident to date landed on the firm as a commercial remedy. Deloitte Australia partially refunded a government report containing AI-fabricated content, and nobody was struck off . Our 2027 case assumes the penalty migrates from the firm's invoice to the individual's licence, and in consulting that has not happened yet.
Buyers are running the other way. The share of clients paying outside consultants for AI support rose from 81 per cent to 88 per cent into 2026 . Strategy, which is board-facing work, was one of only two service lines to shrink in 2023, the opposite of our story, where the Buyer cuts rented staff first . And Gartner found that 80 per cent of large companies using autonomous AI tools had cut staff without getting the returns they expected . The cheap in-house team our Buyer builds is, on today's evidence, neither cheap nor quick.
The measured time saving is half the figure we assumed, and uneven. Our timeline assumes delivery hours fall sixty per cent. The best-documented figures are McKinsey's own claim that Lilli, its in-house AI tool, saves up to 30 per cent of the time spent searching and synthesising , and a field experiment on 758 consultants, published in 2023 and run with GPT-4, that measured 25.1 per cent faster completion . In that experiment the consultants using AI were 19 per cent less likely to be correct on tasks outside what the model is actually good at, which the study calls its competence frontier, and 23 per cent worse on the judgement-heavy task , . Real, concealable compression exists, and the same experiment rated the AI-assisted work 40 per cent higher in quality . Our figure still runs ahead of the evidence, and the gains are jagged: AI helps a lot on some tasks and hurts on others. That 2023 experiment is the best measurement in the pack and the oldest, and the pack contains no later measurement of the same design.
The hiring collapse is uneven across the profession. Counting graduates and apprentices together, KPMG cut its 2024 UK intake by about a third , , Deloitte by 22 per cent, EY by 11 and PwC by 9 . A second report, counting graduate schemes alone, puts Deloitte at 18 per cent, EY at 11 and PwC at 6 . On either count only KPMG and Deloitte cut by fifteen per cent or more in one year, as our Firm does in its first. BCG grew, with 7 per cent revenue growth in 2025, its twenty-second consecutive year, and headcount up to 33,500 , . Our Firm is a composite of the steepest cutters. If the political clock fires, it fires on an uneven, contested labour signal.
An affirmative AI indemnity already exists. Ending A stands on nobody credibly indemnifying model output. Since April 2025 a Lloyd's-market product co-developed by Armilla and the underwriter Chaucer has affirmatively covered hallucinations, meaning AI inventing false information that reads as fact, along with model drift, meaning a model's answers getting less reliable over time, and related AI failures . Microsoft's Copilot Copyright Commitment already indemnifies commercial customers for copyright claims, and says nothing about whether the output is correct . The limits are small today, but cover of the kind that would collapse Ending A is already on sale.
So what - for a leadership team
Your own team is betting the firm on assumptions it has not said aloud.
You are treating clients value trust and accountability and clients will keep paying the old price as one claim. They are two claims. A reputation for judgement keeps clients coming to you. It does not keep them paying the old price for work they now know is cheap to make. , So ask, out loud: if buyers confine us to the thirty per cent that needs a signature, is that thirty per cent a viable business at our cost base?
You cannot starve your new capability to protect today's revenue and also have it ready when you need it. Ask who inside is actually allowed to make the new unit win, and whose pay that win comes out of.
Your binding constraint over the next two years may be the capital your own captive has to hold against AI-heavy work, and the price of the reinsurance above it, long before any regulation arrives. , Has anyone worked out your insurance and your compliance cost as a bill that rises with how aggressively you use AI? Is your chief financial officer already the real brake on your strategy, and do you know it?
The fastest erosion may come from your own customers building the capability in-house, ahead of any competitor. How much of your current revenue is work a client could do itself the day it has the tooling and one good hire?
Cutting your intake of juniors gets booked as an efficiency move. It may also be a decision to trade a 2035 shortage of senior people for 2027 margin, on the assumption that senior judgement can be bought later. , Where do your 2035 leaders come from if you saw off the ladder below them now?
You answer those by putting four to eight of your own people in a room and forcing the disagreement into the open. Give somebody in that room the job of attacking the comfortable answer.
That is what the Workshop is. Half a day. You bring the decision that is keeping you up. We bring the scenario and the red team. We run your assumptions until the weak ones fall over, in a room on a Tuesday, before the market runs the same test over three years.
And if a workshop is not the right tool, we will tell you. Sometimes the honest answer is that you have an execution problem, or a data problem, or a nerve problem, and half a day of structured disagreement would just be theatre. The simulation on this page is the free version. The value is in running it on your own decision, before the fork closes.
- Dr Dan Epstein, The Long Game Project
Changelog
v0.17, corrected 19 October 2026. The Firm's 2028 graduate intake now reads 42 for every 100 it hired in 2025, a cut of fifty-eight per cent. That is the value the story's own record holds for 2028. The text had said 60 and forty per cent. The method notes are gone, and who played and how is in the editorial policy. No other number changed.
v0.17 - 12 October 2026. The report now speaks as "we", meaning The Long Game Project. The opening answers the subtitle's question in plain words, including how the insurer moves the saving. The roles, the players and the dates now agree across the three readings. The graduate-intake figures are reconciled with their two news sources, the Challenger's price is set against the small-deal fee it implies, and the text now states the values the price and insurance charts plot and admits the contrary real evidence where each first appears. A new section says where the real record stops, and names the three signs to watch from the first draft. The Firm's graduate intake now follows one trail at every depth: 85, 60 and 33 for every 100 it hired in 2025. The price and insurance charts now name what each bar measures, because the story's number and the real one rest on different bases. The chart of buyers paying for more AI help moved to the section it belongs to, and the insurance rise, the intake cut and the chart labels now agree at every depth. The seams list more of the numbers we chose. This changelog is shorter. No checked fact changed.
v0.16 - 12 October 2026. The method is called a simulation throughout.
v0.15 - 12 October 2026. Claims checked against their sources. One claim with no source was cut: that every managing partner is privately asking the opening question. No number, marker or heading changed.
v0.14 - 8 September 2026. A Big Four partner read the essay, and the argument changed where the partner showed it was weak. The Firm's traits are now split by type of firm, the insurance step rests on a captive insurer (an assumption, listed in the seams), the first step explains how the intake cut lifts margin, and the small-deal price is marked as the floor of the market. No number or marker changed.
v0.13 - 6 September 2026. Plain-language rewrite, so a reader from another field can follow it. Every technical term is defined where it first appears, and metaphor and closing riddles are cut. No number, date, claim or marker changed.
v0.12 - 5 September 2026. Editorial pass. Two markers moved off a sentence their sources did not support. The reading convention moved to the opening, and the eight places where the evidence cuts against the story got their own heading.
v0.6 to v0.11 - 31 July to 26 August 2026. Readability and design passes. Semicolons and invented icons went, and the figures were redrawn as standard charts of checked data.
v0.4 and v0.5 - 20 July 2026. First interactive edition.
v0.3 - 19 July 2026. Hardening after a hostile review before publication. Citations were re-pinned so that an invented motive never reads as a checked claim about a named firm, and an eighth place where the evidence cuts against the story was added.
v0.2 - 6 July 2026. Every real-world claim tied to a checked evidence pack of 58 claims. About a third of the proposed citations failed the check and were cut. One correction: an insurer named as the 2027 price-setter had no supporting evidence, so the text now says "the carriers".
v0.1 - 1 July 2026. First public draft. One story, five roles. The endings are ranked by our judgement, not by evidence.
Sources
Every marker in the text resolves to a verified claim below - 54 of the pack's 58 claims are cited. A claim marked * is true with a recorded caveat. In the claims below, MBB means McKinsey, Bain and BCG, the three largest strategy firms, PI means professional indemnity, and "the draft" means an earlier draft of this report. Full provenance, verbatim quotes and retrieval dates live in the evidence pack (throat-to-choke-v1).
| Id | Verified claim | Source |
|---|---|---|
| s-001 | KPMG announced in July 2023 a multibillion-dollar commitment to Microsoft cloud and AI services over five years, with the alliance framed around a potential incremental growth opportunity for KPMG of over US$12 billion. | Microsoft (official news release), 2023-07-11 |
| s-002 | EY's September 2023 EY.ai launch carried a US$1.4b headline investment that the firm's own release described in the past tense as spend that had already been made, giving direct primary-source support to the pattern of headline 'AI investment' figures being substantially retrospective rather than new incremental commitments. | EY (official news release), 2023-09-13 |
| s-003 | EY's release attributes the US$1.4b to embedding AI into pre-existing proprietary EY technologies and to previously completed technology acquisitions, i.e. money already sunk into existing platforms rather than fresh frontier-AI spend. | EY (official news release), 2023-09-13 |
| s-005 | PwC became OpenAI's first ChatGPT Enterprise reseller and its largest enterprise customer in May 2024, rolling the tool out to more than 100,000 US and UK employees on top of a separately announced $1 billion three-year AI roadmap. | CIO Dive, 2024-05-29 |
| s-006 | Deloitte's October 2025 alliance with Anthropic makes Claude available to more than 470,000 Deloitte people, Anthropic's largest enterprise AI deployment to date, with 15,000 professionals to be certified on Claude and a dedicated Claude Center of Excellence. | Anthropic (official announcement), 2025-10-06 |
| s-007 | KPMG made the steepest UK graduate intake cut of the Big Four, trimming its graduate cohort from 1,399 in 2023 to 942, a reduction of roughly a third that sits inside the draft's fifteen-to-forty per cent cut band. | City AM, 2025-06-23 |
| s-008 | Deloitte cut its UK graduate scheme by 18 per cent (inside the draft's 15-40 per cent band) while EY and PwC cut 11 per cent and six per cent respectively (below the band), showing the cut band fits some but not all Big Four firms. | City AM, 2025-06-23 |
| s-009 | McKinsey says it now generates about a quarter of its global fees from outcomes-based pricing rather than time-and-materials billing - the firm's own figure, relayed via late-2025 reporting - supporting the draft's shift-to-outcome-pricing assertion at roughly 25 per cent of fees for the leading MBB firm. | Hunt Scanlon Media, 2025-12-08 |
| s-010 | Unity Advisory explicitly positions itself as AI-led rather than legacy-based and targets firms with revenues between £500 million and £1.5 billion, meaning the flagship AI-native challenger is aimed at the upper mid-market, not the sub-$30m deal segment in the draft archetype. | Consultancy.uk, undated |
| s-011 | AI-native challenger DiligenceSquared (YC Fall 2025, $5m seed) sells commercial due diligence analysis for about US$50,000 against the US$500,000-$1 million PE firms pay McKinsey, Bain or BCG for the equivalent study - roughly 5-10% of the incumbent price, which undercuts the draft's assertion that challengers price at 25-40% rather than a tenth. | TechCrunch, 2026-03-05 |
| s-012 | DiligenceSquared matches the 'small senior team plus AI, no pyramid' archetype: its three co-founders are an ex-Blackstone principal, a seven-year BCG private-equity-practice veteran and an ex-Google engineer, with senior human consultants verifying the AI-produced output rather than a junior leverage pyramid. | TechCrunch, 2026-03-05 |
| s-013 | McKinsey itself now runs 40,000 human employees alongside 25,000 AI agents, grounding the 'people plus AI' cost-base framing at incumbent scale and giving the challenger's 'three people plus AI instead of thirty plus AI' contrast a real incumbent-side benchmark. | Fortune (via Yahoo Finance), undated |
| s-014 | The closest documented head-to-head of AI-assisted versus unaided consultant output, the Harvard Business School/BCG field experiment on 758 BCG consultants, found those given GPT-4 completed on average 12.2 per cent more consulting tasks and finished them 25.1 per cent quicker than consultants working without AI. | The Harvard Crimson, 2023-10-13 |
| s-015 | The same BCG/Harvard experiment found consultants using AI on tasks outside the model's competence frontier were 19 per cent less likely to produce correct solutions than consultants without AI, which undercuts any flat claim that off-the-shelf tooling reproduces 60-70 per cent of a consulting deliverable across the board - the reproduction rate is task-dependent, not uniform. | The Harvard Crimson, 2023-10-13 |
| s-016 | In BCG's own writeup of the experiment, consultants who used GPT-4 on the business problem-solving task performed 23 per cent worse than those doing the task without it, showing that AI reproduction of consultant work collapses on judgement-heavy analysis even while it lifts performance on creative and drafting tasks. | BCG Henderson Institute, 2023-09-21 |
| s-017 | Enterprise buyers are still paying external consultants around AI rather than fully insourcing: Source Global Research found 81 per cent of clients in Q2 2025 said they had paid consultants for AI support in the previous 12 months, a share that rose to 88 per cent by Q1 2026. | Source Global Research, 2026-03-20 |
| s-018 | Evidence on which consulting categories buyers cut first is mixed: Source Global Research data shows strategy - board-facing work, not staff augmentation - was one of only two service lines to shrink in 2023, contracting 2 per cent, which complicates the draft's claim that buyers kill research-grunt spend first and leave board-facing work untouched. | Source Global Research, 2024-09-24 |
| s-019 | The main structural vehicle for enterprise insourcing of analytical and technology capability - India's global capability centres - now employs over 1.9 million professionals on global engineering, AI and product mandates, with roughly 50 new centres opened in the first half of CY2025 alone, supporting the claim that enterprises are building internal delivery capability at scale (though as large offshore centres, not four-to-eight-person cells). | Zinnov, 2026-01-08 |
| s-020 | The 'near-zero marginal cost' insourcing framing is undercut by buyer-side outcome data: a Gartner poll of 350 companies with at least USD 1 billion revenue found 80 per cent of those using autonomous AI tools had cut staff but were not achieving the returns they expected, with Gartner warning that workforce reductions create budget room but not return. | IT Pro, 2026-05-07 |
| s-021 | The 2025-26 professional indemnity market is in a soft cycle, not a hardening one - Australian specialist broker Bellrock reported in July 2025 that PI discounts of 5 to 15 per cent were available for firms with good claims histories amid abundant insurer appetite, so the draft's annual-repricing mechanic is real but the current direction is rate reductions with capacity entering, and the 15-40% AI repricing band found no support in any fetched source. | Bellrock Advisory, 2025-07 |
| s-022 | By July 2025 the 'silent AI' debate was already a named concept in UK insurance-law commentary, defined as the risk of policies that do not explicitly state whether AI risks are covered or excluded, directly grounding the draft's 'quiet argument' about whether silent AI was already inside existing wordings. | Browne Jacobson, 2025-07-31 |
| s-023 | Norton Rose Fulbright's regulatory blog stated in May 2026 that the silent-cyber dynamic of unintended, unpriced coverage lurking in traditional policy lines is now repeating with AI, supporting the draft's framing that the silent-AI argument follows the silent-cyber playbook. | Norton Rose Fulbright (Global Regulation Tomorrow), 2026-05-21 |
| s-024 | In October 2025 Deloitte's Australian member firm agreed to partially refund a US$290,000 report for the Department of Employment and Workplace Relations after AI-generated errors including fabricated references and a fabricated federal court quote were exposed, with the sanction landing on the firm as a commercial refund rather than on any licensed individual. | Fortune, 2025-10-07 |
| s-025 | The UK FRC's March 2026 generative and agentic AI guidance for audit keeps regulatory accountability for AI tool use and audit quality with firms and named Responsible Individuals, making 'the human remains responsible' a fair one-line summary of current professional-body and audit-regulator guidance. | ICAS, 2026-03-30 |
| s-026 | KPMG UK cut its graduate and apprentice intake by around 33% in 2024 and reduced its UK workforce by 1,266 employees, the sharpest entry-level pullback among the Big Four. | Scottish Financial News, 2025-02-03 |
| s-027 | Deloitte UK's 2024 graduate and apprentice intake fell to 2,150 (617 fewer, roughly 22% down year-on-year), with EY down about 11% to 1,600 and PwC down about 9% to 1,450. | Scottish Financial News, 2025-02-03 |
| s-028 | Microsoft and LinkedIn's 2024 Work Trend Index (31,000 knowledge workers, 31 countries) found 52% of people who use AI at work are reluctant to admit using it for their most important tasks, and 53% worry it makes them look replaceable. | Microsoft WorkLab, 2024-05-08 |
| s-029 | Slack's Fall 2024 Workforce Index (17,372 desk workers, 15 countries) found 48% of desk workers would be uncomfortable admitting to their manager that they used AI for common workplace tasks, with feeling it is cheating (47%) and fear of seeming less competent (46%) or lazy (46%) as top reasons. | Slack (Salesforce), 2024-11-12 |
| s-030 | In the Harvard Business School/BCG field experiment on 758 BCG consultants (Dell'Acqua et al., 'Navigating the Jagged Technological Frontier'), consultants using GPT-4 completed tasks 25.1% more quickly, finished 12.2% more tasks, and produced 40% higher quality results, establishing the magnitude of task-time compression available to conceal. | One Useful Thing (Ethan Mollick, study co-author), 2023-09-16 |
| s-031 | Ten top-25 US business schools now charge more than a quarter of a million dollars in total two-year MBA cost (Columbia $269,829, Stanford GSB $266,517), with the top-25 average at $230,901, so the draft's $150-200k debt-service framing is if anything conservative on total programme cost. | Poets&Quants, 2025-08-27 |
| s-032 | For the MBA Class of 2024, 22 of 24 leading MBA programmes reported a decline in the share of graduates accepting consulting offers (Tuck highest at 44%, HBS 18%, Stanford GSB 14%), with 14 of 23 programmes at five-year lows in consulting acceptances. | Clear Admit, 2025-11-17 |
| s-033 | McKinsey is weighing a few thousand job cuts staggered over 18-24 months, with reductions of up to 10 per cent discussed for some non-client-facing teams, on a headcount a professional-services recruiter says is already down about 25% from peak at roughly 36,000. | The Register, 2025-12-16 |
| s-034 | MBO Partners' 2025 State of Independence study counts more than 72 million Americans working independently, with a record 5.6 million independent professionals earning over $100,000 a year (up 19% from 2024), evidence for the fractional and boutique exodus of high-earning professionals. | MBO Partners, 2025-09-09 |
| s-035 | Deloitte reported aggregate global revenue of US$70.5 billion for FY2025 (year ended 31 May 2025), up 4.8% in local currency on FY2024. | Deloitte Global, 2025-09-30 |
| s-036 | EY's FY2025 global revenue reached US$53.2 billion on 4.0% growth, versus Deloitte's US$70.5 billion on 4.8%. | Consulting.us (Consultancy.org network), 2025-10-30 |
| s-037 | Thomson Reuters' Future of Professionals research finds 34% of professionals are using AI tools their organisation has not approved - evidence that bottom-up AI adoption in professional services is running ahead of firm governance. | Thomson Reuters, undated |
| s-039 | UK professional indemnity broker guidance benchmarks PI premiums at 0.25% up to 5% of fee income or annual turnover depending on risk factors and market competition, so the draft's 15-40% PI 'repricing' should be framed as a multiple applied within this 0.25-5%-of-fees band, not as a share of fees itself. | Professional Indemnity Insurance Brokers (professionalindemnity.co.uk), 2025-01-17 |
| s-040✱ | The best-documented incumbent delivery-hour reduction figure is McKinsey's claim that its Lilli platform saves consultants up to 30 per cent of the time once spent searching and synthesising information, roughly half the draft's 60 per cent assumption and limited to research/synthesis tasks rather than whole engagements. | Entrepreneur, 2026-06-11 |
| s-041✱ | Unity Advisory, the ex-EY/PwC-led 'next-generation CFO advisory firm' launched in June 2025, is backed by an initial equity line of up to US$300 million from Warburg Pincus and targets private equity-backed and other upper mid-market businesses, making it the most capitalised AI-native challenger to the Big Four. | Warburg Pincus, 2025-06-30 |
| s-042✱ | DiligenceSquared's own launch positioning is '80% of the insight at 5% of the cost' versus US$500K-$1M traditional projects that take 2-4 weeks, showing a live AI-native DD challenger deliberately anchoring near 5% of incumbent price - evidence against the draft's claim that credible challengers price at 25-40% rather than a tenth. | Y Combinator, undated |
| s-043✱ | Bridgetown Research, founded by an ex-McKinsey strategist and backed by a US$19 million Series A co-led by Accel and Lightspeed, claims to produce an initial due-diligence analysis in 24 hours with inputs from hundreds of respondents - a live fixed-clock DD offer far faster than the draft's ten-business-day archetype, with no published money-back guarantee. | TechCrunch, 2025-02-26 |
| s-044✱ | Trade coverage of consulting's productisation argues that independents and small firms now lose on demonstrable, structured, traceable delivery rather than raw quality - only partially supporting the draft's line that boutiques fail on distribution rather than delivery, and reframing the binding constraint as provable method plus visibility. | Consultancy.uk, undated |
| s-045✱ | The nearest mainstream precedent to a money-back guarantee is incumbents' shift off the billable hour: PwC's productised 'PwC One' pricing is expected to move to subscription and outcomes-based models, so outcome-linked fees are entering mainstream advisory even though no named AI-native DD challenger publishes a money-back guarantee. | Consultancy.uk, undated |
| s-046✱ | US federal courts in 2025 imposed sanctions for AI-fabricated citations on individual licensed attorneys - including disqualification from the case and state bar referral in Johnson v. Dunn - rather than on firms as entities, confirming that in law the liability incidents to date have attached to the licensed human. | Sterne Kessler, undated |
| s-047✱ | Since April 2025 an affirmative AI liability insurance product co-developed by Armilla and Lloyd's underwriter Chaucer has covered hallucinations, model drift and legal liability arising from AI underperformance, so the draft's Ending A assumption that no insurer credibly indemnifies model output is already false at current (small-limit) scale - and the documented named market pioneers are Chaucer/Armilla and Munich Re rather than Chubb, for which no AI-PI evidence was found. | Chaucer Group, 2025-04-22 |
| s-048✱ | The broadest AI vendor indemnity in force, Microsoft's Copilot Copyright Commitment announced September 2023, commits Microsoft to defend commercial customers and pay judgments only for third-party copyright infringement claims and only where guardrails and content filters were used, accepting no liability for the accuracy or professional adequacy of outputs - consistent with the draft's premise that labs and platforms do not indemnify professional-grade work. | Microsoft, 2023-09-07 |
| s-049✱ | The same 2024 Work Trend Index found 78% of AI users bring their own AI tools to work rather than waiting for employer-sanctioned ones, meaning most workplace AI use happens outside official measurement, which helps explain why firm-level AI ROI lags individual gains. | Microsoft WorkLab, 2024-05-08 |
| s-050✱ | Ethan Mollick's 'secret cyborgs' work reports survey evidence that over half of generative AI users at work use the technology without telling anyone at least some of the time, grounding the claim that employees pace AI-accelerated delivery to look like the old cadence. | One Useful Thing (Ethan Mollick), 2023-03-08 |
| s-053✱ | PwC's FY2025 gross global revenue was US$56.9 billion, up 2.7% in local currency for the year ended 30 June 2025 - the slowest growth of the Big 4. | Consulting.us (Consultancy.org network), 2025-10-30 |
| s-054✱ | PwC's global headcount fell by 5,600 in FY2025 to 364,000 people - direct evidence that Big-4 headcount is already contracting, not just slowing. | Consulting.us (Consultancy.org network), 2025-10-30 |
| s-055✱ | BCG (the only MBB firm that publishes revenue) reported US$14.4 billion for 2025, up 7% from US$13.5 billion in 2024 - its 22nd consecutive year of growth. | Boston Consulting Group via PR Newswire, 2026-04-23 |
| s-056✱ | BCG's global workforce grew to 33,500 employees in 2025 - so unlike the Big 4, at least one MBB firm is still adding headcount. | Boston Consulting Group via PR Newswire, 2026-04-23 |
| s-057✱ | Thomson Reuters' 2025 Future of Professionals report estimates AI will free up nearly 240 hours per legal professional per year (up from 200 in 2024), worth about US$19,000 per professional annually - a sourced lower bound on how many billable-type hours per head AI is already collapsing. | Thomson Reuters Institute, 2025-06-26 |
| s-058 | By May 2026 broker Gallagher reported that three carriers were seeking regulatory approval to exclude AI-driven losses from their professional indemnity and commercial general liability policies, meaning AI exclusions in PI wordings are an observed market movement rather than a hypothetical. | Gallagher, 2026-05 |