Why Almost No Law Firm Will Actually Do This

For general counsel and legal operations evaluating whether a firm's technology story is real.

Every argument about efficiency in legal services ends in the same place. Fine, but firms can change. If hourly billing stops the saving reaching the client, a firm can move to fixed fees, build the technology, and pass it on.

True. Some will. Almost none will as partnerships, and the ones that manage it mostly stop being partnerships first.

What it costs, from someone who can afford it

In May 2026 the Financial Times reported that Kirkland & Ellis had set aside $500 million to build its own AI platform, expecting to spend over $100 million in 2026 and hundreds of millions more across three to four years. Around 250 of its lawyers contributed to the design. More than 180 technology professionals are building it. The firm owns the output, and the outside companies involved cannot sell it to anyone else.

The useful figure is not the $500 million. It is the annual one.

Kirkland's revenue was $10.6 billion in 2025. The $100 million it expects to spend this year is a little under 1 percent of that. The full commitment, spread across three to four years, works out at roughly $125 to $165 million a year, so 1.2 to 1.6 percent. Which is where Kirkland's chair Jon Ballis put it when he told Bloomberg Law that a firm should invest around 1 percent of revenue in new initiatives.

So the headline number is enormous because the revenue is enormous. The ratio is ordinary. Any firm could adopt it.

Now apply that same ratio to the firms doing the work most amenable to automation. AIPLA's 2023 economic survey, based on 157 firm responses, puts median total billings for a US IP firm at $2.2 million. One to one and a half percent of that is $22,000 to $33,000 a year.

That is not a shortfall, it is an order of magnitude, and it does not close by being clever. The cost of building a portfolio management system does not fall to a hundredth because your firm is a hundredth the size.

A partnership has no mechanism for this

Even where the money exists, the structure fights it.

Profit per equity partner at Kirkland was $11.1 million in 2025. An investment at that scale is absorbed without any individual partner noticing it in their own distribution.

Run the same decision in a twenty-partner firm. The money comes from the pool that pays the partners, in the same year, and there is no debt facility, no outside investor, and no retained earnings mechanism built for a multi-year project. Every unit invested is a unit not distributed, and it is distributed annually while the payback runs over years.

So the question in the room is not whether the technology is a good idea. It is whether twenty people, some of them five years from retiring, will each accept a smaller number this year for a return that arrives after several of them have left.

That is a governance problem wearing a technology costume.

The part nobody budgets for

Even with the money and the mandate, there is a second cost that appears in no business case.

You cannot automate a process nobody wrote down.

Most firms do not have a documented workflow for prosecution. They have experienced people who know what to do next, which is not the same thing and cannot be handed to a system. Before anything gets built, somebody has to define what actually happens, in what order, with what decision points, across every jurisdiction the firm files in. That work is slow, generates no revenue while it happens, and is done by the people whose billable hours the firm can least afford to lose.

Buying a tool does not solve it. A licensed platform imposes somebody else's process, which is why so much legal software is bought and quietly abandoned. The tool was fine. The firm had no process to put in it.

Why Kirkland is not a counterexample

The obvious objection is that Kirkland is doing it, so the barrier cannot be that high.

Two things. It is funded from a revenue base almost no other firm has, at a ratio its own chair describes as what scale permits. And the work their platform serves is not the work under discussion. Capturing institutional knowledge for complex litigation and transactional matters turns efficiency into throughput on matters that are already value-priced or capacity-constrained. Nobody's fee shrinks.

Automating renewals and prosecution is different, because there the hours are the fee. The incentive problem is not evenly distributed across legal services. It is worst in high-volume, procedural, jurisdictionally repetitive work, which is exactly the work most amenable to automation.

Which is the squeeze. The firms with the capital to build have the least reason to let it reduce a fee. The firms whose work would benefit most have no mechanism to fund it.

What the firms that do it actually do

They stop being partnerships.

Over the last three years, three European IP firms have taken private equity investment and said technology was the reason.

A London-founded IP network operating across sixteen jurisdictions sold a minority stake to a private equity house. Its chief executive told JUVE Patent that the amount you now have to invest, and the expertise needed to get value from technology, data and process, means classical partnerships are not always fit for purpose.

A Dutch firm merged with a French one and took investment from a Dutch private equity firm, abandoning its partnership structure in the process. Its managing partner said the investment needed to become a pan-European player would not have been financially possible without it.

A century-old Dutch firm merged into the first network explicitly to gain resources for technology, its managing partner citing the scale of investment required in AI over the coming decade.

Alongside those, an ASX-listed holding company has run a network of patent and trademark firms since 2014, citing technology-driven efficiency in its listing case. An AIM-listed UK firm has made fifteen acquisitions in ten years, including a trademark boutique in 2025.

Not one of these is a partnership funding a build out of distributed profit. Every one changed its ownership structure first, and several said so on the record.

The same pricing model, opposite incentives

Ballis was explicit about pricing in the same FT interview. People talk about the evolution of the billable hour, he said. The firm already does a number of matters on value-based pricing, that trend will accelerate, and they are looking forward to leaning into it.

Worth reading carefully, because value-based pricing does two different things depending on who is doing it.

Charge for the value delivered rather than the cost of delivering it, and a firm that becomes dramatically faster keeps the entire gain. For the highest-grossing firm in the world, at $11.1 million of profit per equity partner, that is what it looks like. There is no share left to win. Getting faster and pricing on outcome converts efficiency into margin, and any sensible business in that position would look forward to it.

Now run the same model in a firm trying to take work off incumbents. The efficiency has to go into the price, because the price is what wins the work. Holding it back to pad margin means not growing. That is not virtue, it is arithmetic. In a market where the incumbent bills by the hour, a lower and more predictable number is the whole argument.

So the pricing model tells you little on its own. What tells you something is whether the firm needs to grow. A dominant firm captures the efficiency. A firm taking share spends it. Same fixed fee, opposite direction, and the difference is competitive position rather than character.

What to actually ask

If a firm tells you it has built technology that makes it more efficient, four questions separate the real from the rehearsed.

What did you build, and what did you license? A firm running on a generic platform with services layered on top has bought a subscription, not built a capability.

How was it funded? Not the amount, the mechanism. Retained profit, outside investment, or a parent company. The answer tells you whether the commitment survives a bad year, and whether the partners had to agree to it.

What did you have to write down first? If they cannot describe the process definition work, the automation is probably shallow. This is what separates a workflow from a wrapper.

And what happens to my price when it gets better? The only one that matters. If the answer is nothing, the efficiency is theirs.

Where we land

We are not an exception to any of this. We are another instance of it.

Digip took venture capital, built the software, and now runs on its own cashflow. That is the same structural answer the firms above arrived at. We just started there instead of converting, which meant we never had partners to persuade and never had to choose between a distribution and a release.

That is the honest reason we can price the way we do, and it is why we do not expect many firms to follow by the same route. Admitted counsel does the legal work in every jurisdiction, ours or one of the 15 or more local counsel firms in our network, across more than 190 jurisdictions. The software handles everything around that, which is what lets us charge a subscription plus a fixed all-in fee rather than metering hours.

Every efficiency we find goes into that number, because that is how a small firm takes work from much larger ones. Not generosity. It is what growth costs when the incumbent bills by the hour.

The four questions above work on us as well as on anyone. The last one especially.