Posted by Eduard Grigoyan, general counsel of Legal Futures Associate Codified Strategy
Grigoryan: Who will own the relationship?
On 23 June, Legora changed how it charges.
The Swedish-built platform, formerly Leya, now used within Linklaters, Cleary Gottlieb, Goodwin and a long list of firms most readers will recognise, announced that its most capable product was moving away from the flat per-seat licence fee.
In its place comes consumption-based pricing. You pay for the work the AI actually does (presumably plus a margin), and every run can be attributed to the matter that prompted it.
What has actually changed
Until now, Legora sold what most legal AI vendors sold: an annual licence, charged per user, offering effectively unlimited use. The price varied by firm, by size, by region, and was negotiated behind a non-disclosure agreement.
Heavy users and light users paid the same, and the cost sat in the firm’s overhead, invisible to the client.
Consumption-based pricing breaks that apart. Firms will now pay in proportion to what the tool does and a real-time dashboard will track that consumption by organisation, each user and, critically, each project.
In Legora’s own words, every run can be tied back to the matter, giving firms cost transparency for each client.
The move to consumption-based pricing applies only to Agent Pro, Legora’s top tier. The lighter version stays fixed fee, and Legora has not published any prices for this.
The talk of ‘tokens recharged to clients’ is not Legora’s language. It is the market spelling out what matter-by-matter pricing makes possible.
The vendor’s logic is not unreasonable
An AI agent that plans, drafts, checks its own work and iterates consumes wildly different amounts of compute on a one-page letter than on a 200-document review. A flat fee per seat cannot absorb that variance: the vendor either overcharges the light user or loses money on the heavy one.
Clay, Cursor and the model providers themselves have all moved this way. Legora is following a trending path, not inventing one.
The problem is what the model quietly does to two things a firm cannot afford to lose: its margin and its client.
The questions a pricing deck will not answer
Firms will need to start with the bill. Once a cost can be tied to a matter, the temptation is to recharge it to the client, like a search fee or counsel’s fees.
But first a firm has to answer something its IT consultant cannot and its vendor has no reason to raise – is AI consumption a disbursement or an overhead dressed up as one?
Software has always been overhead, part of what your fee already buys. Recharge it wrongly as a client-facing line and you are not facing a commercial nuisance, but a conduct and accounts problem under the Solicitors Regulation Authority’s rules.
The answer may come from the e-disclosure space, which has charged based on the volume of data for years. But that model only holds because the cost sits outside the firm and is visible to the client – two things AI running on your own systems cannot offer.
The opposite choice carries its own risk. Absorb the cost, and a variable, usage-driven bill sits badly on fixed-fee work where the firm carries the risk – the matter that runs hot costs more to service, with nothing extra to recover.
Either way, the decision belongs in your engagement letters and your pricing, settled before the meter is running, not after the vendor’s first invoice lands.
Whose client is it?
Now the part that should occupy a managing partner’s thoughts.
Picture the client’s invoice 12 months from now. There, every month, is a line that reads ‘Legora’. The client starts to notice it. Then the client starts to associate the work – the answer, the draft, the review – with the name on the line.
Slowly, in the client’s mind, the tool becomes the thing doing the lawyering and the firm becomes the thing operating the tool.
This is particularly the case as Legora, along with Harvey, have made a very deliberate point around building an ecosystem in which clients and lawyers can collaborate. They want to become the ecosystem, in the same way that Microsoft did for many businesses.
Legora is not a company that will quietly disappear. It passed $100m in annual recurring revenue this year, serves more than 1,000 customers, and was reported to be growing faster than OpenAI. Its most recent funding round put a multi-billion-dollar valuation on the business.
A company at that scale, with the client’s name already sitting on its own dashboard, has every commercial incentive eventually to ask the question the line item plants for it: why is the client paying the firm to operate the tool, when it could pay for the answer directly?
I have written before [1] that you would not buy legal advice from someone who has never practised law. The mirror of that question is now being put to the profession from the other direction.
The honest answer is that the value of a lawyer is in judgement, accountability and a relationship with a client whose business the lawyer actually understands. No tool, however capable, carries that.
But a firm that lets the vendor become the visible brand on the bill is handing over, line by line, the one asset it can never buy back: the relationship.
This is not an IT decision
Whether to adopt consumption-based AI, how to characterise the cost, what goes in the retainer, how to keep the client relationship the firm’s own – none of this is surfaced by a 30-minute demo or a vendor’s pricing slide.
They are not infrastructure questions and they are not purely commercial ones. They are questions for someone who has billed a client, reconciled a client account, and read the SRA rules under pressure.
Consumption-based pricing is coming to legal AI, whether the profession is ready or not, and Legora will not be the last to move. And when clients can see exactly what the AI did on their matter, the conversation about what the firm charges for that matter won’t be far behind.
The question will not be whether the meter runs – for all but the very few who build their own infrastructure, it will. It is whether, when the client reads the invoice, the margin and the relationship are still yours.
Codified Strategy works with law firms on technology strategy, AI implementation and practice management. Our consultants are qualified lawyers. They have done the job and understand your perspective whilst having deep knowledge of the legal technology market.