Even last week, Anthropic's General Counsel Jeff Bleich told the American Bar Association that AI will "destroy" the billable hour. Bleich is right that the billable hour is under pressure and firms need to adapt. But I don’t think AI is going to destroy it.

The question isn't whether AI will kill the billable hour. It won't. But AI is doing something no previous technology managed to do: make legal work predictable. That changes the pricing conversation entirely.

I explored some of these ideas in a recent conversation with The Legal Wire. This piece builds on that and outlines what’s really happening and where the conversation needs to go.

This conversation is older than you think

Bleich isn’t the first person to make this call. The death of the billable hour has been predicted in every decade since it became the profession's dominant model in the 1970s. Document automation was supposed to kill it. E-discovery technology was supposed to kill it. Legal research software was supposed to kill it. Each wave of technology produced the same prediction, the same conference panels, the same think pieces, and the same result. The billable hour endured.

It’s not because the billable hour is invincible. It’s because the billable hour is largely misunderstood, as are the technological innovations that have previously called for its demise.

Let's be clear about what the billable hour actually is

Most people predicting the demise of the billable hour talk about it as a pricing mechanism. That’s too simplistic. As legal industry veteran Mary O'Carroll describes it: the billable hour isn't just how firms charge clients—it's the operating system of the entire enterprise. It shapes how associates are evaluated, how partner compensation is structured, how performance is measured at every level of the firm. You can't swap in a new billing model without touching everything else built on top of it.

But there's a second dimension that even O'Carroll's framing understates. At its core, the billable hour is a risk-shifting mechanism, which is why it has proven so durable.

When a firm begins a matter, it rarely knows how complex that matter will become. A contract review that looks straightforward on day one can become a six-month negotiation. A regulatory inquiry that seems contained can expand into a multi-jurisdictional investigation. The firm cannot predict this at the outset, and neither can the client. The billable hour resolves this uncertainty by passing the risk to the client: if the matter becomes more complex, the client pays more. If it stays simple, the client pays less. The firm is protected; the client absorbs the variance.

This isn’t a bug, it’s a feature. Clients have accepted this arrangement for decades because the alternative — asking a firm to commit to a fixed price for genuinely unpredictable work — exposes firms to potential losses. A firm that prices a $1 million fixed fee and then spends $2 million delivering the work has made a $1 million mistake.

Understanding this changes the conversation about what it will take to replace it. Any alternative model must answer the same question the billable hour answers: who bears the risk when work is unpredictable? No alternatives will scale unless they can solve the risk allocation problem first.

Alternative pricing is already underway

The key point that tends to get lost in the debate is: law firms are already doing a version of value-based pricing. They just don't call it that.

Think about how work actually gets priced today. A corporate client issues an RFP. Firms submit pitches. Each firm offers an estimate — "we think this matter will cost approximately X." Clients select based partly on that number. That estimate is not a suggestion. In practice, it is a de facto pricing commitment, even when the underlying arrangement is technically hourly. Win work by saying it will cost $1 million and then bill $2 million, and you will have a very unhappy client — one who selected you on the basis of a number you did not honor. The relationship suffers. The next panel review goes differently.

The billable hour is not a taxi meter that runs until you get out. It is closer to an Uber ride: there is a quote at the outset, a rough expectation of cost, and a negotiated relationship built around that expectation. Firms already compete on value. The mechanisms are just less transparent than they should be.

There's a structural reason alternative fee discussions stall even when both sides claim to want them: shadow billing. Clients frequently demand hourly tracking on fixed-fee matters to benchmark whether they're getting a good deal. If the firm's hourly equivalent comes in well below the flat rate, the client feels overcharged — even when the outcome is identical. Smart firms know this, and calibrate fixed fees so the shadow billing lands just above the flat rate. The result is a fixed fee that is, in practice, an hourly arrangement in disguise. The gap between firms offering alternative arrangements and those arrangements actually changing how work gets priced is larger than adoption statistics suggest. It's not resistance — it's a trust problem neither side has fully solved.

According to Best Law Firms' 2025 survey of over 4,800 U.S. firms, 72% now offer alternative fee arrangements, rising to 90% among firms with more than 50 lawyers.

The gap between where legal is today and where it needs to go is narrower than the "death of the billable hour" framing implies. The infrastructure for change already exists. What's missing isn’t willingness—it's the right conditions to accelerate.

Lessons from the Big Four

If you want to understand where legal is heading, look at professional services industries that have already been through this transition.

Accounting is the clearest parallel. The Big Four once billed almost entirely by the hour. Over the last 20 years, that shifted almost entirely. Routine, repeatable work — tax compliance, audit engagements, recurring advisory — moved to fixed fees. Complex, bespoke, high judgement work still commands hourly rates, but the pattern is consistent: standard, repeatable work shifts to fixed pricing first, where competitive pressure is highest and clients have the most leverage.

Legal is behind that curve, but following the same trajectory. In 5-10 years, the legal industry will likely look like accounting: a diverse mix of billing models, with hourly rates concentrated where they're genuinely justified and fixed fees standard everywhere else. Not a wholesale replacement, a rebalancing. Some practice areas will get there before others. Contract review, due diligence, and legal research will shift first. Trial work and complex high-stakes strategy will be last. That's the model working as designed.

AI is an accelerant, not a detonator

Most of the conversation about AI in legal has focused on efficiency: faster research, quicker drafting, fewer hours spent on repetitive tasks. Efficiency matters. But it’s only part of the story, and arguably less consequential.

The deeper impact of AI is predictability.

Efficiency means doing the same work faster. If a task that took ten hours now takes one, the firm is more efficient. But under hourly billing, that efficiency is a problem: the firm has just reduced its own revenue by 90% on that task. The client benefits but the firm absorbs the loss. This is the central tension that makes firms reluctant to adopt new technology, and why the ‘AI will destroy the billable hour’ framing triggers defensiveness rather than action.

Predictability changes the equation entirely. When AI makes work repeatable and consistent, firms gain something they have never had before: certainty over what the work will actually cost to deliver. Historically, fixed fees were difficult for law firms due to uncertainty. AI reduces that uncertainty for specific task categories — early-stage due diligence, first-pass document review, contract analysis, regulatory research — by making those tasks consistent and measurable. When firms can predict the time and effort required, they can price the work accurately.

This shift is already playing out, quietly, through outside counsel guidelines. Any partner who has received one knows the experience: a client sends a document, unilaterally, detailing what they will and will not pay for. That list has been growing for years — legal research, proofreading, first-year associate work, travel time — items that clients have concluded they should not pay for.

AI accelerates this dynamic. As clients develop a clearer understanding of what AI can do, and at what cost, they become less willing to pay human hourly rates for work that AI can handle. The firms that have not adopted AI aren't just falling behind technologically. They're absorbing costs they can no longer recover. They still have to do the research, the proofreading, the diligence, they just simply can't bill for them. Their cost structure expands while their billable universe contracts. In other words, they get left in the dust.

There is a scenario worth noting that complicates the AI acceleration narrative: in practice areas where uncertainty remains high, AI may not lead to fixed fees at all. It may instead lead to higher partner hourly rates. AI will primarily reduce time spent on lower-level tasks — the work historically handled by junior associates — and is unlikely to meaningfully reduce partner hours, since partners already delegate automatable work downward. As firms lose the associate billings that once drove profit, they face a choice: charge directly for AI usage, raise partner rates, or find a new pricing model entirely. The total cost to the client still decreases, the savings from eliminating associate hours offset the rate increase. But the billable hour at the partner level does not disappear. In some practices, it may actually become more valuable as the volume of hours shrinks and the judgment behind each one becomes more concentrated.

The biggest obstacle to new pricing models isn’t firm resistance. It’s that most firms lack the understanding of what AI can actually do, practice area by practice area, to know where the pricing opportunity lies. Broad AI adoption is different from deep AI integration. Until firms have built genuine capability in specific areas and automated meaningful workflows, pricing conversations remain abstract. The path to pricing innovation runs through deep AI integration, not just procurement.

The real moment of change isn't negotiation, it's competition

Here’s the reality of how pricing innovation happens in professional services: it often doesn’t happen with existing client relationships. It starts with competitive pitches for new work.

Renegotiating pricing with an existing client is difficult. The relationship has a history. There are expectations on both sides. Introducing a materially different pricing model mid-relationship requires both parties to agree that the old model was wrong, which creates friction even when the new model is clearly better. Most firms avoid that conversation unless forced into it.

New pitches are different. When a firm is competing for work, it can offer a different pricing structure without disrupting an existing relationship. It can use that structure as a competitive differentiator. AI-enabled pricing innovation will spread through competition first, and existing client renegotiation second. The firms that understand this will move early on new pitches, build a track record with new structures, and then have a data-driven case to bring to existing clients. The firms that wait will find themselves in a weaker negotiating position when it arrives.

Pricing innovation happens through competition, not conversation.

Four models are filling the gap. One doesn't exist yet.

The firms navigating this transition well aren’t replacing the billable hour with a single alternative. They are building fluency across a broader set of models, and matching the pricing structure to the nature of the work.

Fixed fees are the most established alternative, and already standard for predictable, repeatable work: contract review, routine due diligence, straightforward compliance matters. The client gets cost certainty. The firm gets margin upside if it executes efficiently, and AI makes efficient execution more achievable. As AI compresses delivery costs on these tasks, fixed fees become more attractive for firms to offer, not less, because the cost basis drops while client willingness to pay a flat rate holds.

Capped fees and phased billing have gained traction as a client-facing trust mechanism, particularly for matters with definable stages. Litigation is the natural fit: cap the investigation phase, cap filing, price trial separately. Clients get budget predictability; firms retain some protection against scope uncertainty. The honest caveat: capped fees are asymmetric. If the matter runs long, the firm absorbs the overrun. If AI makes the firm dramatically more efficient, the client captures most of that gain. They work as a competitive differentiator and a way to build trust with new clients — but they are a bridge, not a destination.

Portfolio pricing is a fixed fee covering a defined set of similar legal matters over a set period — employment disputes, IP filings, routine litigation — rather than billing each one hourly. For clients it delivers budget certainty on predictable, recurring work. For firms it rewards efficiency: the faster and better you handle the volume, the better your margin. The incentive alignment is cleaner than capped fees — both sides benefit when the firm operates well. And beyond the economics, portfolio arrangements build something strategically valuable: the institutional familiarity and data integration that comes from being deeply embedded in a client's recurring legal work. A firm that knows a client's patterns, risk profile, and history is much harder to replace than one receiving discrete instructions and returning discrete work product.

And then there's the pricing model that’s just on the horizon: The billable token.

As AI agents take on more of the substantive work of a matter, the unit of billing shifts. It is no longer purely the attorney's hour. It is the compute behind the output: the research synthesized, the documents reviewed, the drafts generated, the analysis produced. A markup model for AI work that mirrors the economics of the billable hour, but built for a world where the associate doing the work is an agent, not a person. The firm pays for compute. It marks that compute up — embedding it within a legal services layer that adds judgment, supervision, and accountability — and charges the client accordingly. The incentive alignment mirrors the billable hour: more work done correlates with more value delivered and more revenue generated. The economics, however, are fundamentally different.

We have been thinking a lot about the billable token at Legora and will explore it fully in a follow-on piece. But the direction is worth naming now, because the firms thinking seriously about it today are the ones who will have pricing power when it becomes standard practice.

Collaboration infrastructure makes this all possible

There is a dimension of this conversation that gets less attention than it deserves: the infrastructure of collaboration.

Trust but verify isn’t a slogan. It’s the first principle of every AI-driven legal workflow. AI doesn’t remove the need for lawyers to check each other's reasoning, inspect outputs, and build shared institutional knowledge. If anything, it makes that process more important. When AI is generating research, drafting documents, and running analysis, human review becomes the mechanism of quality control.

But collaboration also needs to extend beyond the walls of the firm. Today, the relationship between law firms and clients is structurally inefficient. Work moves through email threads, document versions, and phone calls with no shared visibility into status, scope, or progress. That fragmentation makes pricing conversations harder. When a client can’t see what work is being done or how, conversations about cost and scope become adversarial. When the workflow is visible, structured, and shared, those conversations become collaborative.

We built Portal to reduce this friction — creating a shared interface through which work moves between firms and clients with transparency on both sides. Clients who can see the work can have an honest conversation about how to pay for it. That transparency is not just an operational improvement. It’s the foundation on which pricing trust gets built.

The future isn't monolithic, and that's the point

The framing of ‘billable hour versus value-based billing’ is a false binary. The future of legal billing isn't a single model. It's a more diverse, practice-specific ecosystem, and the firms that thrive will be fluent across all of it.

More predictable workflows mean more predictable cost structures. Better cost visibility means more confidence in fixed-fee pricing. The uncertainty that made the billable hour a rational hedge against complexity diminishes as AI makes the work more standardized and repeatable.

The billable hour will remain for high-stakes, complex matters where time spent really is the best proxy for value delivered. It stops being the default answer to everything else.

The death of the billable hour has been predicted for decades. It's not going to die. It’s a slow, uneven, practice-by-practice evolution that’s already well underway.