A Sydney law firm we spoke with last month billed a contract review at four hours, the same review Claude Fable 5 helped their associate finish in fifty minutes. The client paid for four hours anyway, because the retainer was still time based. The partner running that file admitted the arrangement was hard to defend once she thought about it properly. That gap between what AI-assisted work actually costs to produce and what firms still bill for it is now the central pricing question in Australian professional services.
The hourly billing problem AI creates
Hourly billing was built for a world where speed and value moved together: a faster lawyer, accountant or consultant delivered more billable hours, and the firm earned more for it. Claude and similar tools break that link. A grant application that used to take a consultant twelve hours might now take three, once drafting, research and formatting run through Claude Code or a Claude-built workflow. If the firm still bills $220 an hour, the invoice drops from roughly $2,640 to $660, and the client notices straight away.
That is not a hypothetical. We have watched three Australian advisory clients hit this exact wall in 2026: delivery time fell by half or more, invoices did not, and clients started asking pointed questions about what they were actually paying for. Two of those firms lost a repeat client over it before they changed how they priced.
Three pricing models worth testing
Most firms land on one of three approaches once they stop billing by the hour for AI-assisted work.
Fixed-fee packages: price the outcome, not the hours. A tax structuring review might move from an hourly estimate to a flat $3,500, regardless of whether Claude cuts the drafting time in half.
Value-based pricing: price against what the work is worth to the client, not what it cost to produce. A compliance audit that prevents a $50,000 regulatory penalty is worth more than its production hours suggest.
Hybrid retainer plus fixed fee: a monthly retainer covers ongoing Claude-assisted advisory work, with fixed fees layered on top for discrete deliverables like reports, filings or board papers.
Fixed fees tend to work best for services with predictable scope, things like BAS preparation, standard contract reviews or onboarding packs. Value-based pricing suits higher-stakes advisory work where the outcome matters more than the method. The hybrid model is what most firms actually settle on, because it protects recurring revenue while still letting you charge properly for one-off deliverables that used to eat a full week.
What clients actually want disclosed
Clients are less worried about whether you used Claude than whether they're being charged fairly for the outcome. Most don't ask about tools at all until pricing feels out of step with turnaround time. The firms handling this well treat AI use as a normal part of delivery, not something to hide or oversell, and they say so plainly in their engagement letters: what's automated, what's reviewed by a person, and who's accountable if something goes wrong.
For firms handling personal information under the Privacy Act, that disclosure matters for a second reason: clients are entitled to know when their data is being processed by an AI system, and engagement letters should say so rather than bury it in a schedule nobody reads. Accountants and lawyers also carry professional body obligations around supervision, CPA Australia and the state Law Societies both expect a qualified person to review AI-assisted output before it goes to a client, and that review time belongs in your price even when the drafting time has shrunk.
A practical framework for setting your price
Start by timing the AI-assisted version of a service against the pre-AI version for the same deliverable, across five or six recent files. If a $4,500 fixed-fee engagement now takes your team eight hours instead of twenty, don't simply cut the price to match the new hours. Price against the three things clients actually pay for: the judgement applied to the output, the risk you're carrying if it's wrong, and the speed they're getting.
A Melbourne accounting firm we advised settled on lifting its fixed fee by roughly 15% for AI-assisted advisory work, while cutting turnaround from three weeks to five days, and told clients exactly why in the fee proposal. Nobody pushed back, because the proposal led with the outcome the client was getting, not the hours behind it.
When not to reprice yet
If you have only just started using Claude for client work, hold your existing rates for a full billing cycle before touching pricing. You need real data on how much faster delivery actually gets across a representative set of files, not a guess based on one impressive draft. Repricing too early, before you understand where a human still needs to spend real time checking the output, tends to cost more in redone work than it saves in fee cuts.
If your invoices are shrinking faster than your margins can absorb, that's a pricing conversation, not an AI problem. We help Australian professional services firms work out where Claude actually changes the cost to serve, and how to reprice around it without a client backlash. Book a session if you want a second set of eyes on your fee structure before the next round of engagement letters goes out.
