Hey friends:
It’s been a busy summer for me! And also a busy summer for all the innovation players in the legal ecosystem too, I suppose. I thought I’d take a few minutes out of this Sunday afternoon to highlight 3 stories from the ecosystem that I’m watching carefully:
1. Harvey making moves to enable future growth
Earlier this week, Harvey launched a free Legal Engineering certification program. Right after, they announced the release of their second generation model, Harvey II, that leverages their first proprietary, post-trained model. These two developments appear somewhat unrelated but they indicate that Harvey—despite its massive commercial success—is looking ahead to protect future growth.
Having your own proprietary model reduces reliance on Big AI. Creating Tenet, the underlying model behind Harvey II, probably took a lot of time and energy to execute. It likely required a herculean effort, including leaning on teams of contract attorneys from Mercor, SnorkelAI, and others to train the models. The upshot is that Harvey is taking a step away from being “just an AI-wrapper” which should enable future independent growth.
Separately, the creation of a certification program usually means the company is thinking about retention—a key enabler of massive future growth. It’s a classic playbook from SaaS (e.g. Salesforce) and legal (e.g. Relativity); I remember one of my earliest initiatives at Logikcull was to create our own certification program to reduce churn, enable future growth, etc. Here, Harvey is doing something slightly different. By targeting the highly undefined LE role (as I shared here) Harvey isn’t locking in users to their platform but instead acquiring mindshare and brand association for this growing category of roles.
2. How much will PE pay for a law firm?
Also this week, it was announced that Wood Smith—a California insurance defense firm—signed a LOI to sell to private equity via a MSO structure. If you’ve been paying attention, the trend of law firms becoming increasingly (partially) owned by non-lawyer entities has accelerated. What makes this particular deal notable is that (1) it’s not a plaintiffs’ firm and (2) for (I think) the first time the numbers behind the deal have been made public.
By itself this story is notable, but against the backdrop of increasing examples of investors/owners getting around rules governing law firm ownership—it’s even more interesting. Non-lawyer ownership is becoming increasingly common. In the past, many of these PE/MSO transactions involved personal injury firms; now we’re starting to see Amlaw firms and now a large regional defense firm joining the mix.
If this trend becomes and established model, it’ll lead to a permanent split between (A) firm operations on the one hand, and (B) actual delivery on the other. This can be highly effective in many types of law practices, especially those that involve relatively standardized work (maybe that’s why personal injury firms have been early adopters). However we’ll see how this plays out when it comes to more judgment-intensive bespoke work.1
3. Increasing legitimacy of AI-native law firms
This summer was full of announcements by existing and new AI-native law firms that tells us this new category is here to stay. Norm AI announced a $120m round at $1.2 billion valuation and Manifest OS announced a $60m round on a $750m valuation. And not only are new players emerging, we’re also hearing veteran establishment lawyers joining or starting these types of firms. (If Bloomberg writes about it, you know it’s a real trend!)
The one tidbit of news I found most interesting though came out of Crosby when they announced recently that they’re seeking to provide professional liability insurance for their AI agents. This could be a critical move to unlock AI-native firms. I’ve always thought that AI-native firm lawyers would be hesitant to truly delegate tasks to AI, and instead spend a lot of time checking over the work. Which would work against the whole point of using AI for speed & efficiency.
However, if you can insure against those mistakes—then perhaps the need to double check goes away.2 Especially if we find that AI’s error rate is lower than human review. Unfortunately, establishing insurance does seem to suggest that the AI is engaging in the practice of law (which is a big no-no) so there’s admittedly a large looming regulatory cloud hanging over it all. However, unlike other similarly regulation-ambiguous artifacts (e.g. MSOs, AZ ABS) insurance might be welcomed by clients, especially if it lead to faster review times & flat fees.3
Conclusion
It seems that there is a huge amount of investor interest in the legal space. That seems to be driving a lot of the resources pouring into our little corner of the world—and is creating all sorts of new, interesting structures & entities designed to get around rules designed for an older era. Time will tell whether technology and capital will lead to better outcomes for clients. Fun times ahead!
I’d be very curious how things are structured if a traditional Biglaw firm sells to PE. Seems like you cannot separate the institution from the lawyers practicing law that’s highly complex.
Insurance is fascinating to me because it commercializes a fundamental principle I learned in contracts during 1L year, which was “you can decide to comply with the contract, or breach and pay damages.” I’d never before realized that you actually never have to do what the contract says, if you’re willing to pay to make your counterparty whole. (Whether or not that’s a desirable outcome is a separate question).
With insurance that protects against AI mistakes, you, as a lawyer might be freed up to rely on AI more heavily to gain speed/efficiencies—and then if something goes wrong, have someone else pay the financial cost or damages incurred from the error. Having this type of insurance could massively accelerate sanctioned AI usage among risk-averse lawyers. I have a lot of thoughts on this topic that I might articulate in a future post.
Although the ethics and professional responsibility rules seem archaic, they serve a pretty critical role in protecting clients. And my sense is that adjudicators can be flexible on the application of those rules—but only when that would help a future client or the public. (Admittedly this might be more of a hunch than a proven conclusion so I’m open to other opinions).
But if someone creates a structure or artifact to get around the rules solely to enrich private equity or non-lawyer owners, I suspect there will be higher barriers to adoption. Contrast that with how remote Zoom-based solo practices are viewed; they don’t always follow the letter of the law for bar requirements (e.g. need to have a physical office in the state) but regulators do not seem to be going after them—probably because they help increase access to justice.
Again, all this is just my sense, not based on anything concrete.