Hiring outside counsel used to be a function of handshakes and law school connections. To some extent, it still is. But over the past several years, the in-house legal function went through the same “efficiency expert” transformation that hit other industries decades earlier. Now there’s benchmarking, and ROI calculations, and arcane billing guidelines to make sure outside counsel expend a lot of time and energy to clarify exactly how they spend their time and energy.
Clients have a more defensible case to justify their budgets to the board, firms previously locked out of the old lawyer network can find themselves winning business, and all this new efficiency drastically drives down legal costs.
Well, two out of three ain’t bad.
But one Chief Legal Officer just announced on LinkedIn that his company is going to take this cold efficiency drive to another level by turning over law firm hiring to AI.
Oh…
Revolut is a company whose “global mission is for every person and business to do all things money — spending, saving, investing, borrowing, managing, and more — in just a few taps.” That’s not particularly clear, but the company takes a customer’s salary and puts it in an account and issues spending cards. That would make you think they’re a bank, though the website clarifies in small print that “Revolut is not a bank.” It’s been called “disruptive,” and whenever I see that term I like to replace it with “bullshit.” Which doesn’t necessarily mean it can’t be a fine service! But it’s “bullshit” as in an investor-friendly ersatz version of something that already exists. So it’s sort of a bullshit bank — probably with some AI element to money management. Ah, there it is!
Anyway, Tom Hambrett, Revolut’s Chief Legal Officer, formerly of Herbert Smith Freehills, announced that the company is scrapping the traditional law firm panel in favor of an AI-assisted system:
We’re replacing our static panel with Revolut Partners – a dynamic, performance-based ecosystem assessed against metrics that actually matter & competitive best in class rates, and of course how they engage with our products.
No one partner’s position in the starting lineup is guaranteed. Firms will be reviewed quarterly – and we’ll make changes when the model tells us to.
I’m sorry, you’re right. You said this was a bet-the-company litigation and I hired a firm advertising on late night cable. Would you like me to run a new search for bankruptcy counsel?
All right, that might be too harsh. Hambrett describes the process further, suggesting that the algorithm will be fed a curated and vetted population of firms:
That bench is real. We’re building a pool of firms ready to step up the moment a first-choice partner falls short. Underperformance means: poor client management, unmanaged scope creep, weak billing practices, a lack of responsiveness, or direct feedback from our lawyers that advice quality isn’t where it needs to be.
A “pool” of firms, eh? One might even, with the benefit of a thesaurus, call that a “panel.” They say it’s not “static” — whatever that means — but at the end of the day, the client is vetting a group of firms and entering them into the system. That sure sounds like a panel. But a panel with its own branding as “Revolut Partners,” which sounds like an idea that the AI came up with.
We’re also building in-house AI tools to help lawyers run RFPs, pre-select firms for specific instructions and scrutinise advice & invoices. Firms can no longer solely rely on soft relationship touch points to keep their place in the squad.
Relationships still matter. But performance wins you minutes on the pitch. The firms that make our shortlist will be the ones that consistently deliver the right outcomes, at the right pace, with the right commercial discipline.
When you get past the ad copy of “a dynamic, performance-based ecosystem assessed against metrics that actually matter,” this just sounds like some tech-enhanced benchmarking for a panel of law firms, which is… fine? In-house legal departments perform some version of this already, with or without AI tools. The biggest difference is that Hambrett promises that “we’ll make changes when the model tells us to.” I’m a little skeptical that the company will really turn its legal fortunes over to a “GPTsus take the wheel” philosophy.
Roll on Friday caught this conversation and surveyed the responses:
Leanne Murray, a real estate lawyer at Priestleys, agreed that while relationships are still important, businesses “need to ensure” that “KPI’s are being met.”
“No room for complacency! Golden goose panel systems that a business is stuck in for years is changing,” Murray added. “Your panel firms should always behave and perform in a way that is an extension of your business.”
Bart Lieben, an IP lawyer also concurred with Hambrett: “Transparency on metrics should be the default, not something the client has to ask for.”
All totally fair reactions, though also nothing that requires turning over decision power to a bot. Not only can AI assist in-house teams in evaluating outside counsel, it’s probably reckless at this point NOT to use AI to analyze data. But there’s a difference between that and outsourcing judgment. Nobody sane objects to a CLO getting a dashboard. Telling the world that a technology prone to making up fake cases will start making decisions could alienate some firms.
That’s one lawyer cited by RoF explained succinctly:
“Adverse selection could be an issue. The firms with genuine market leverage will simply opt out. They have less demanding clients and simply might not need this. What remains is a competitive race among firms that couldn’t afford not to participate (on a desperate part of the market for some reason), which is the opposite of what this model seems to focus on.”
Yeah. Firms with genuine market leverage — the ones with steady, less demanding, faster-paying clients — can simply decline to enter a quarterly knife fight judged by an AI bot giving a thumbs up or down from the imperial box. The best law firms out there don’t have time for loyalty resetting every ninety days. They’re in it for the long haul and want their books to reflect ongoing commitments. What’s then left auditioning for the bench is the slice of the market that can’t afford not to. That model is optimized to reward whoever shows up, but not to make the good ones show up.
Honestly, everything about this post is — absent the startup jargon — a just good legal department management right up to “we’ll make changes when the model tells us to.” Maybe he meant to that as “we’ll make changes when the model’s results dictate that’s the right response.” If so, that’s fine. But something tells me a bank-that’s-not-a-bank-but-with-AI probably meant it quite literally.
Uh-oh: Company decides to let AI rate and pick its law firms [Roll on Friday]
Joe Patrice is a senior editor at Above the Law and co-host of Thinking Like A Lawyer. Feel free to email any tips, questions, or comments. Follow him on Twitter or Bluesky if you’re interested in law, politics, and a healthy dose of college sports news.