Two kinds of law firm are being built right now. One starts with AI and hires lawyers around it. The other starts with lawyers and buys AI around them. Both will tell you they are the future. Only one of them is cheaper this quarter.
The AI-native lead is real
An AI-native firm has no legacy to defend. No comp plan built on leverage, no realization targets that assume a certain number of associate hours, no workflow bolted together in 1997 that someone now has to migrate. It can price on scope instead of time, because its cost to deliver was never denominated in hours in the first place.
That is a genuine structural advantage, and it shows up as price. When an AI-native firm quotes well under the incumbent and still holds margin, the incumbent can match the number only by explaining what changed about the work. That is an uncomfortable conversation to have with a client who paid last year's rate.
Add the attention. Right now the excitement, the capital, and a fair amount of the talent interest are pointed at the native side. It looks like a rout.
The catch-up is easier than it looks
Here is the problem with that read. Technology is the most purchasable part of a law firm. Models are rented by the seat. Whatever an AI-native firm has assembled on top of frontier models, some version of it becomes licensable to everyone within a year or two, because the vendors selling it have every incentive to sell it twice.
So the pure technology moat is shallow. It buys a lead measured in quarters, not decades, and the length of that lead is set less by the technology than by how fast the incumbent decides to move.
What incumbents own is harder to buy
The incumbent's advantages are the unglamorous, unbuyable kind:
- A general counsel who will not be second-guessed by a board for hiring the known name.
- Panel positions, referral networks, and decades of matter history.
- Bench depth for the matter that actually threatens the company.
- A balance sheet, insurance, and a track record of surviving a bad year.
Brand and distribution are a tougher moat than a tech stack. You can buy a model. You cannot buy forty years of being the firm that got called.
These firms are slow, and slow is a real liability here, because the retrofit is hard in ways the demos never show: partner incentives point the wrong way, and the operational work that AI would absorb was never written down anywhere. That is the test. But slow is not the same as still, and there is a good chance the incumbents win it.
The move to watch: buying the brand
If credibility is the binding constraint for AI-native firms, the fastest way through it is not marketing. It is hiring. Expect the native side to pay up for marquee partners, the recognizable names with portable client lists, precisely to short-circuit the trust and brand gap that would otherwise take a decade to close. In this market, a lateral is the only distribution you can acquire outright.
And that experiment cuts both ways, which is what makes it worth watching. If the laterals land and their clients follow, the incumbent's moat was never the institution, it was the individuals, and it can be bought one partner at a time. If the clients stay put, the brand was institutional after all, and the native firms have a much longer road than the current excitement suggests.
What actually decides it
Not model access. Both sides end up with the same models. The difference is how much of the firm's own operating work, the intake, conflicts, matter setup, billing, and collections, is standardized enough for AI to run at all.
AI-native firms are ahead there for a boring reason: they designed the process before the volume arrived. Incumbents have the volume and no documented process, which is why buying more AI has not closed the gap for them. Tools accelerate a process. They do not create one.
The good news for the incumbent is that this gap is closable without a two-year documentation project. The process can be captured from how the work is already being done, standardized into one way of doing it, and automated from there.
That second gap is the one Caddi exists to close. Not the judgment you sell, but the operational work around it: Caddi learns how your firm already runs a task, turns that into one standard way of doing it, and then runs it. The incumbent advantage is real. It just needs a business that can move at the speed of the brand.