The Harvey effect: how AI is reshaping law firm pricing

The uncomfortable truth about how legal work gets priced

Winston Weinberg, co-founder of Harvey, made a point recently in the Financial Times that many in the profession know is true but rarely admit in public. When a client pays a law firm £100,000 for a piece of work, they are really paying for a few hours of a senior partner's judgment. But the invoice rarely reflects that. Instead, it gets padded out with hundreds of hours of junior associate research, all billed at rates that obscure what the client is actually buying.

That structure made sense when junior associate time was the only way to get the research done. It makes less sense when AI can compress that work dramatically - and when clients know it.

In-house teams are already doing the maths

Weinberg's observation cuts to the heart of the change underway: in-house legal teams are not going to keep accepting bills for tasks they can now run through AI at a fraction of the cost. Once a general counsel knows that the first pass of a contract review or a due diligence exercise can be produced in minutes rather than days, the conversation with outside counsel shifts. Not necessarily to cutting fees, but to demanding that pricing better reflects the value being delivered, as opposed to the inputs consumed.

This is not a future risk for law firms. It is a present one. And the firms thinking carefully about how to respond are already asking a harder question: if AI compresses the time it takes to do certain tasks, what does the residual matter actually cost, and how should it be priced?

The compression problem - and the opportunity in it

The challenge for law firm pricing teams is that matter economics are changing mid-stream. A transaction or dispute that historically required 400 hours of associate time might now require 150. That is not a loss of revenue if the firm prices intelligently. The partner's judgment - the advice that the client is really paying for - is still worth what it was. But the way to recover that value is no longer to bill junior time at inflated hourly rates.

This is where data becomes critical. To price a matter well in an AI-augmented world, you need to understand which tasks within it are being compressed and by how much. That is not a question most firms can currently answer with any precision. Pricing decisions are still largely based on historical matter data that does not distinguish between tasks that AI will handle and tasks that still require human expertise.

Ayora's data enrichment capability addresses this directly. By analysing matter composition - breaking down the work into its constituent task types - it becomes possible to model how AI adoption changes the time profile of a matter. Firms can see not just what a similar matter cost historically, but what hours are likely to remain once AI handles the compressible parts. That gives pricing professionals an honest foundation for building fixed fees, blended rates, or any other structure that reflects current economics rather than legacy assumptions.

Pricing policy that actually reaches the practitioner

Knowing what a matter should cost is only half the problem. The other half is ensuring that the right pricing approach is applied consistently - across practice groups, across offices, across the hundreds of individual quotes that leave a firm every month.

This is where most pricing functions struggle. A pricing committee can set a policy. CVT can build a model. But when a partner sits down with a client to discuss fees, they are often working from instinct rather than from a principled framework. The result is inconsistency - some matters priced well, others priced to win at the expense of margin, and almost no visibility into which is which until the matter closes.

Ayora's pricing agent changes that dynamic. It takes whatever policy the firm has set - whether that is a pure fixed fee, a blended arrangement combining time and materials with fixed elements, or anything in between - and surfaces it at the moment a quote is being built. Pricing professionals and attorneys alike see the same framework, applied consistently, every time. The partner still makes the final call, but they make it with the right information in front of them rather than relying on memory or instinct.

This matters more as pricing structures become more complex. A blended fee arrangement that combines a fixed element for due diligence with time and materials for negotiations is harder to apply consistently than a straight hourly rate. Without tooling that propagates the policy automatically, the risk of drift is high.

The bridge between practice and business

There is a deeper point here that Weinberg's interview touches on, even if indirectly. The legal industry has historically treated the practice of law and the business of law as separate domains. Partners focus on the work; finance and pricing teams focus on the numbers. That separation is increasingly hard to sustain.

AI forces the question. To price a matter in an AI-augmented world, you need to understand both the legal substance of what is being done and the economic structure of how it should be recovered. Those are not naturally the same conversation, and most tools designed for legal teams are not built to bridge them.

Ayora is designed from the ground up to do exactly that. The user experience is deliberately straightforward - the same kind of intuitive interface that has made tools like Harvey easy for lawyers to adopt. But underneath it, Ayora combines an understanding of how legal matters are structured with economic and commercial analysis skills. It is not a finance-facing templating tool, or a legal tool that gestures at finance. It is built to sit at the intersection of the two.

That matters because the transition Weinberg describes - from hours to outcomes, from inputs to value - requires firms to have pricing conversations that are simultaneously legal and commercial. The partner needs to understand the economic implications of a fixed fee proposal. The pricing professional needs to understand enough about the matter to model it accurately. Ayora gives both of them a shared language and a shared platform.

What this means in practice

The billable hour is not going away, as Weinberg himself acknowledges. But its role is changing. It will increasingly be one pricing mechanism among several, applied where it makes sense and replaced with fixed or blended structures where it does not. For clients, that shift is largely positive - they get more predictability and more confidence that they are paying for value. For firms, the opportunity is to price that value properly rather than letting margin erode through inconsistency and outdated assumptions.

Getting there requires better data, better tooling, and a clearer line of sight between what AI is doing to matter economics and how that flows through to pricing decisions. That is a solvable problem. But it requires firms to start treating pricing as a discipline that sits at the heart of practice management - not as a peripheral function that catches up with the work after it has already been scoped.

The firms that figure that out first will find themselves in a stronger position with clients who are already asking harder questions about what they are paying for. The ones that do not will find those conversations increasingly difficult to navigate.

‍