The AI productivity surplus is up for grabs, and your fee model decides who gets it

Something interesting is happening inside professional services firms right now, and almost nobody at partner level in Australia is talking about it yet.

Generative AI is producing a real, measurable productivity uplift in knowledge work. The research consensus puts it at 20 to 40 percent in construction management, engineering and professional services. Forrester's TEI on Microsoft 365 Copilot at enterprise scale finds 9 hours saved per user per month, translating to a 116 percent three-year ROI on the composite organisation studied. Well-governed Copilot deployments are showing $5,200 to $19,500 per user in annual productivity value.

That productivity gain sits somewhere. It has to. It either stays with the firm (as fatter margin, faster delivery, better utilisation) or it flows to the client (as lower fees, shorter engagements, or both).

Right now, in most firms, it is sitting quietly with the firm. That is about to change, and the change is already visible in the sectors that are further along.

What the numbers say about who is capturing the surplus today

The most direct evidence comes from the legal sector, which is roughly twelve months ahead of consulting on AI adoption.

Thomson Reuters' 2026 AI in Professional Services Report finds that 58 percent of law firms say AI has had no effect on their billing practices, 23 percent say it has increased efficiency without changing billable hours, and only 19 percent say billable hours have actually declined.

Read that carefully. Four out of five law firms deploying AI are keeping the productivity surplus. Their people are working faster and better, and clients are still being billed at the same rates for the same nominal hours. The dividend is going to the firm's P&L.

That will not hold. The same report notes that 72 percent of US law firms now offer alternative fee arrangements, and the pressure is coming from the client side. Thomson Reuters quotes corporate legal departments framing it directly:

Major clients have grown frustrated watching their outside law firms claim they are using AI while bills keep climbing, not falling.

Consulting is behind legal on adoption but ahead on the commercial response. McKinsey shifted 25 percent of its global fees to outcome-based pricing in 2026, and reportedly had to restructure partner compensation because outcome-based pricing makes revenues volatile in ways the traditional pyramid does not know how to absorb. Bain and BCG are following. HFS Research reports clients demanding up to 20 percent price reductions on professional services engagements as AI productivity comes into the open.

The macro story is clear. The productivity surplus that is currently sitting with firms is being noticed, and the client response is moving from grumbling to explicit rate demands and RFP-level alternative-fee requirements.

Why Australia has a specific window right now

Australia is behind on AI transformation at the leadership level. PM-Partners' 2026 survey finds that only 12 percent of Australian leaders say generative AI is already transforming their business, compared to 25 percent globally.

But that headline lags a much sharper reality on the ground. Autodesk's 2026 State of Digital Adoption in the Construction Industry finds 52 percent of Australian construction firms are already using AI or machine-learning tools, up from around 25 percent in 2023. And, critically, 43.8 percent of Australian firms report at-scale adoption, nearly double the global average of 24 percent.

Australia is ranked number two globally on construction technology adoption. Not last. Second.

The Australian construction and property sector is bifurcated. Around half of firms are running AI in production and starting to compound. The other half have not yet begun. Firms in the second group are still selling time.

For consulting firms that sit inside that ecosystem, project management consultancies, cost management practices, engineering consultants, superintendents, that bifurcation is where the fee-model question becomes urgent. If your competitors are one of the 43.8 percent at-scale adopters, they are already producing at a lower unit cost than you. If they figure out how to price for it before you do, they will win the next tender you both bid on.

Three paths, and what each one does to margin

There are only three commercial responses to the productivity surplus. Every consulting firm will end up with one of them by 2028. Most will choose by default rather than by design.

Path one, stay on time and materials. You are on hourly or day rates. AI makes your people faster. On any given engagement, you record fewer hours. The client gets the same output for less spend. The productivity dividend flows to the client. Your revenue per project drops without your headcount or overheads dropping. Margin compresses. This is the default path. It is also, by a wide margin, the worst one.

Path two, move to fixed-fee or capped-stage pricing. You agree the price for a bounded piece of work. Internally, AI lets you deliver it in fewer hours than budgeted. The margin delta between the priced fee and the actual delivery cost stays with you. This is a defensive posture. It captures the surplus but does not change the sales conversation. It works reliably at stage-priced project work (a bid response, a contract review, a document set) and less well at open-ended advisory.

Path three, move to outcome-based pricing. You tie fees to outcomes the client actually cares about. Schedule adherence. Cost variance to budget. On-time practical completion. Fee margin retained on the client side. AI de-risks the bet, because your ability to predict, monitor and adjust outcomes is materially better with an AI-augmented delivery team than without. The productivity surplus stays with you, aligned to the client's own definition of success. This is the McKinsey move, and it is the model that will define the top tier of professional services by 2030.

Each path has honest downsides. Fixed-fee exposes you on scope creep. Outcome-based makes your revenue volatile and requires you to have real confidence in the measurement layer. Time and materials protects you from both, at the cost of your margin base.

There is no perfect answer. But there is a wrong answer, which is not choosing.

What this means for Australian professional services firms in 2026

Here is what I would say to any managing partner or CFO of a mid-market Australian professional services firm right now, given the evidence base.

One, the productivity surplus is real and you are probably already capturing it. Your senior consultants are using AI on tasks that used to take hours and now take minutes. The saved time is showing up as more capacity per person, which shows up as either better utilisation or fatter margin. You may not be measuring it. It is happening.

Two, the client-side conversation is coming faster than you think. The Australian government procurement framework is already recommending BIM and digital technologies on major projects. Corporate procurement teams read the same industry press their overseas peers read. The "why is my invoice still going up when you claim to be using AI" question will hit your inbox within the next twelve months, whether or not the firm has prepared a response.

Three, the window to move your commercial model without pressure is now. Once a major client asks you for a 15 percent AI-productivity rebate, you are negotiating from a defensive position. Move the model before that conversation lands, and you are shaping the market rather than reacting to it.

Four, the shift is not just a pricing question. It is a proposition question. Firms that move to fixed-fee pricing without changing what they sell will still get squeezed. The firms that will hold and grow their margin base are the ones that combine a new commercial model with a new proposition, augmented delivery capability, outcome guarantees, or productised advisory that packages a repeatable AI-enabled capability rather than person-time. Each of these deserves its own conversation. All three are shapes we are seeing in AU firms that are moving well.

The strategic question, in one line

The productivity surplus from AI is a large, real, transferable pool of value. It currently sits with firms because clients have not yet fully noticed. It will move to clients when they do. In between those two states is a short window in which firms with the right commercial model capture it permanently.

The firms that move first and move deliberately will spend the next five years compounding margin on the back of that capture. The firms that move late will spend the same five years managing fee compression.

Which of those two firms is the one you run.

What we do about this at RYB

This is one of the questions that sits at the middle of the strategic advisory work we do with mid-market Australian firms in construction, professional services and property. Not the technology architecture question, which is well understood by now, but the commercial model question, which is not.

If you want to talk through what the productivity surplus looks like in your specific firm, and what commercial model shift you have room to make in FY27, that is a conversation we run regularly. It is a short piece of work, not a big engagement. It leaves you with a decision to make, not a report to file.

Get in touch if it is relevant.

Tom Leyden runs Red Yellow Blue, a strategic AI advisory practice working with mid-market Australian firms. Sources for this piece include the Thomson Reuters 2026 AI in Professional Services Report, McKinsey State of AI (August 2026), Forrester TEI on Microsoft 365 Copilot, HFS Research (via Santiago & Company), Autodesk 2026 State of Digital Adoption in the Construction Industry, and PM-Partners Australia's AI Transformation 2026.