AI Is Replacing Business Models, Not Just Tasks

Ask a law firm CIO whether their AI investment has produced a positive return, and, if they're being honest, most will say no. Dan Safran, who has worked with over 800 law firms across 35 years of practice, has had that conversation more times than he can count. His observation is blunt: nearly every CIO he speaks with, when pressed, admits they are not seeing positive ROI from their current AI investments.

That tension was the starting point for a joint webinar between August and Unbiased Consulting, with Dan joined by Rutvik Rau (CEO, August) and Hayden Enniss (Solutions Architect, August). The pattern they described is familiar to anyone who has watched a firm's AI strategy unfold. A firm runs an evaluation, signs a contract, launches a pilot. Adoption starts strong and fades. A year later the firm runs another evaluation, often for a tool that overlaps with the first one, and adoption fades faster because the attorneys have seen this movie before. By the fourth or fifth limited-use-case pilot, firm-wide AI fatigue sets in. The innovation budget keeps growing while the business case for it keeps shrinking. Eventually someone in firm leadership asks the obvious question: what are we actually getting for all of this?

The honest answer, for most firms, is hours saved. And hours saved is exactly the wrong answer.

The Measurement Problem

Roughly 85 percent of law firm revenue is billed by the hour. In that model, a tool that makes lawyers faster, measured only in time, is a tool that reduces revenue. Firms have built their entire AI business case on a metric that works against their own economics. It's no wonder partners stay skeptical. They understand the math better than the pitch decks do.

This isn't an argument against efficiency. It's an argument that efficiency only becomes valuable when the firm changes what it does with the time. The capacity AI creates has to go somewhere: into more matters, into better work on the same matters, into services priced on value rather than effort. If it goes nowhere, the firm has spent seven figures to bill fewer hours, and the spreadsheet will say so.

The deeper issue is that hours saved is a task-level metric, and task-level thinking is how the industry got here. Drafting assistance, document review, Q&A over a contract. These are real capabilities, but they're the floor of what the technology can do, and a firm that never gets past the floor will never see a return that leadership can act on. AI in legal has spent three years as a solution looking for a problem. The firms starting to pull ahead are the ones that reversed the order.

A Business Model Question Wearing a Technology Costume

Here's a useful test. The most profitable firms in the country were not made profitable by software. Wachtell moved to an almost entirely alternative-fee model years ago, without AI, because the partnership decided that the value of its work was not a function of the hours spent producing it. Consulting firms made the same move decades earlier. Most consulting work is sold on fixed fees, and the margins on that work routinely exceed what AmLaw firms earn, precisely because the firm captures the upside when it delivers efficiently rather than handing it back to the client as a smaller invoice.

None of that required a model release. It required a decision about how the business works.

What AI changes is the cost of making that decision. Fixed-fee pricing has always carried risk: misjudge the scope and you eat the overrun. Productized services have always carried a delivery problem: the economics only work if the routine core of the matter is genuinely repeatable at low cost. Taking on high volumes of small matters has always been a staffing question with an unattractive answer. AI doesn't invent these strategies. It makes them affordable. The firms treating AI as an accelerant toward a business model they already wanted, rather than as a tool purchase, are compressing a twenty-year ambition into five or seven.

That distinction sounds abstract until you look at where returns actually show up. In practice they fall into four places, and none of them is hours.

Revenue. Matters won because business development follow-up actually happened instead of dying in an inbox. Pricing proposed with confidence because the firm can model matter economics before quoting. Services packaged and sold proactively rather than waiting for a client to ask what the firm can do.

Margin. Every firm carries work that gets quietly written off because a partner won't bill the client for it. Reducing write-downs is a direct margin gain, and so is collapsing the cost of the routine core of a matter, the 24-hour task that becomes one hour because a workflow absorbs it.

Capacity. Attorneys carrying more matters without more hours, and back-office headcount staying flat as volume grows. The overlooked half of this is quality. When the repetitive parts of a matter are automated, the time that comes back becomes strategic thinking time, and the answer the client receives gets better. Corporate legal departments will tell you this is what they actually want from outside counsel; they just phrase it as efficiency.

Position. Turnaround time as a differentiator. Client relationships that compound because the firm's accumulated context makes every subsequent matter start further ahead. These gains never appear on a usage dashboard, and they're the ones that decide who wins the next decade.

What This Looks Like on the Ground

Consider the matters firms turn away. Patent prosecution, real estate closings, routine finance and tax work: high volume, low individual value, uneconomical to staff. Declining them has always cost more than the matter itself, because the small matter is how the relationship starts and the relationship is how the large matter arrives. When the routine core of that work is automated, the math flips. The firm stays in the game at a price point that works, and keeps the client.

Or consider the document problem. A production lands at 11:59 PM with analysis due in the morning. The old answer was a staffing scramble or a declined engagement. The new answer is ingesting hundreds of thousands of documents overnight and surfacing the facts, contradictions, and patterns that matter. That's not a faster version of the old workflow. It's a different service, and firms are starting to sell it as one.

Or consider what may be the most undervalued asset in the industry: the firm's own work product. Decades of matters, briefs, and negotiated positions, most of it sitting inert. Turning that history into something queryable means every new matter starts from the firm's best prior thinking instead of a blank page. It's also the thing that makes a platform feel like the firm's own rather than a generic tool, which is where adoption problems quietly go to die.

And then there's the client side, which firms tend to treat as a separate headache when it's really the same problem. Clients are sending mixed signals: some demand AI-driven efficiency without saying what they mean, others forbid AI outright. Underneath both messages, the goalposts haven't moved. Clients hire firms to solve problems with high-quality answers. A firm that can show its clients how the work gets done, in some cases literally, through shared workspaces where the client watches matters move, turns the AI question from a billing negotiation into a reason to send more work. That's a communication challenge, not a structural one, but you can only communicate what you can measure, which brings everything back to the original sin of measuring the wrong thing.

Start With the Problem, Not the Procurement

If there's one habit separating firms that achieve firmwide adoption from firms stuck in perpetual pilots, it's the question they ask at the start. Stuck firms ask which legal AI is best. Moving firms ask which problems, if solved, would change their business, and then work backward to the tooling, the data, and the people the solution requires. The same logic applies to firms that already own Harvey, Copilot, or anything else. The gap is almost never the tool. It's that nobody defined what the tool was for, so nobody can say whether it's working.

This is also why adoption, the metric everyone frets about, is downstream of everything above. Lawyers don't adopt technology because they're told to. They adopt it when someone shows them an outcome that matters to their practice and their compensation. Define the problem, build toward it, measure the result, and the adoption problem largely solves itself.

Where August Fits

August was built around this premise. It's an AI platform designed specifically for law firms, covering the practice of law and the business of law in one system, with a library of prebuilt workflows so firms aren't reinventing every use case from scratch. More importantly, it's deployed the way the problem demands: every firm August works with is paired with a dedicated lawyer and a dedicated engineer who embed with the firm, learn how it actually operates, and build around its specific problems rather than handing over a login and wishing everyone luck.

Together with Unbiased Consulting, that work runs as a loop. Diagnose where the high-impact opportunities are and build the business case. Design the workflows and agree on the yardstick for success before anything ships. Deploy against the firm's existing systems. Measure against the benchmarks everyone signed up for, then iterate. It takes both halves: the wider lens on strategy, pricing, and change management, and the team on the ground building and shipping every week.

This piece draws on a live discussion we hosted with Dan Safran, President and CEO of Unbiased Consulting, alongside August's Rutvik Rau and Hayden Enniss. A white paper expanding on these ideas, including a practical list of where to start on both the practice and operations side, is coming soon. To get it, or to talk through what this looks like at your firm, book a meeting with a member of our team.