Who gets invited back

How innovation makes its way through the legal ecosystem. Spoiler alert: It's not about the technology, it's about trust & accountability

The shift is hard to ignore. Tasks that used to take lawyers hours can now be completed in minutes. The question of whether AI will affect legal work feels largely settled. We’re even seeing new AI-native law firms show up to the field. All this excitement is taking place against the backdrop of the massive bombshell recently when Anthropic announced its move to legal.

These big announcements show no signs of stopping any time soon. And it’s reaching all corners of the ecosystem, which prompted me to write a two part piece on AI and its intersection with the ALSP world last summer (see here and here). Today I’d like to make the case that while AI certainly has had a big impact—it hasn’t changed how legal decision makers adopt innovation.

Which is through trust & accountability.

1. AI Has Already Had A Huge Impact

For years, commentators debated whether AI would meaningfully change legal work. Not any more. The mechanics of how tasks get completed have already shifted. It’s all led to massive growth and incredible valuations for the top players in the space. Just last week Harvey announced that it was raising yet another round of funding at an incredible $11 billion valuation, prompting me to share this niche tongue-in-cheek post that surprisingly received heavy engagement from tech/VC communities:

It’s not just Harvey though. It’s the steady stream of new startups backed by top VCs, all seeming to raise $5-$10M out of the gate. AI-native firms are the most recent signal of this change. They aren’t just layering tools onto legacy processes; they are designing their operating models around the assumption that intelligence is embedded internally from the start. From Nikki Shaver:

Are AI-native and AI-first firms simply another iteration of boutique experimentation, or are they early manifestations of the kind of disruptive force Clayton Christensen described in The Innovator’s Dilemma? When viewed through that lens, these firms look less like curiosities and more like early fabric-change signals—indicators that the underlying architecture of the legal market may be beginning to shift in ways incumbent firms are structurally ill-prepared to respond to.

Nikki points out that there are at least 20 of these new AI firms out there; it also appears that there are 3 in the most recent Y Combinator cohort alone. (Richard Tromans,, who calls them “NewMod” startups, wrote a great summary of the YC phenomenon here) From Richard:

As AL explored last year, NewMods are certain to make more of an impact this year as the realities of combining fixed fees, super-structured workflows, curated data, lots of AI, as well as experienced lawyers, all in one package, starts to filter through the market. If it were just AI alone, then naturally you would need a legal team to leverage that tech and data. But, with lawyers as part of the offering, then we have a whole different ball game.

That’s a significant development. All this suggests that legal work can indeed be structured differently and potentially delivered with far fewer human bottlenecks than before.

But here’s my point: These changing org structures is just the first step towards change. Faster production—even if promised and delivered to the client—does not automatically translate into widespread adoption. Innovation does not move through the legal ecosystem the way it does in other industries. They move through relationships, risk thresholds, budget cycles, and organizational politics.

AI may be changing how the work gets done. It has not changed how legal decision makers decide where it’s used.

2. Distribution in Legal Is Earned, Not Announced

In many industries, distributing innovation is a function of visibility and scale. Raise a big round from a Tier 1 VC, build a shiny new brand, hire proven go-to-market executives, and land lots of new customer accounts. This is a proven formula.

This approach isn’t quite as effective in a trust-based ecosystem like legal.

In legal, distribution is less about announcements and more about credibility. It almost always starts narrowly, perhaps a single engagement, which only expands if the experience builds confidence. Not merely confidence in the vendor brand, but confidence in how the work was handled when stakes were real (if small). I wrote about this a few weeks ago in a post:

New logos and revenue are positive signs, but increased usage & wallet share over time are probably far more relevant. My experience in the ALSP world suggests that the initial buy decision provides a mixed signal. For example: A legal department may take the first step of “buying” an engagement attorney who’s great at spotting legal issues—but ultimately adds little value to the organization. As it turns out, it’s not enough to be able to merely give good legal answers; you have to also be embedded within the organizational context you operate in.

Growth rarely comes from the first signed contract with the customer. It comes from being invited back. A matter goes well, or even just predictably. Communication is clear. Escalations are handled thoughtfully. Judgment aligns with the client’s risk tolerance. Months later, another matter surfaces. Then a slightly larger one. Over time, scope expands not because of a feature set, but because of familiarity and trust.

That is how wallet share grows in institutional markets: incrementally, through repetition.

This is why distribution in legal compounds quietly. It runs through relationships, clear & constant communication of feedback, and the day-to-day rhythms of legal teams. It is shaped as much by organizational politics and internal alignment as by technical capability. Press releases and announcements move faster than trust. But trust, once established, creates permission—permission to handle more work, more sensitive matters, and eventually more consequential decisions.

And that permission is what durable distribution looks like.¹

3. Legal Institutions Embrace Innovation Through People

Legal departments and law firms don’t innovate in the abstract. They adopt new processes and technologies through people they already trust. You need someone credible to stand behind it. Or “internal champions” in sales parlance. When the innovation is dramatic and has significant change management implications for the organization, the champion needs something more.

They need a partner on the outside they trust & who has deep accountability.

Contrast this with sales or marketing tech. You can hire a junior sales rep to schedule & sell software over the phone. I remember at my first legal tech startup, our VP of Sales told me that he regularly purchases software he learns about from people cold calling him. I was shocked. Because that went against my own (lawyer-trained) instincts about how to buy software.

I think it’s also because if you mess up sales or marketing tech, it’s not a big deal. In those worlds you have to move fast and break things, you work and develop through iteration. So the stakes are lower up front—you can be imperfect in selecting tools or innovation, and still be ok in the end.

Legal work couldn’t be more different. Not only is there little room for error—the lawyers themselves are viewed as shock absorbers for all types of problems. When a lawyer is asked for his/her opinion on the matter, it’s not just the substance that’s relevant—it’s also their credibility; the fact that they are reputable practitioner with extensive experience. Or as Kevin Roose once told me, they are “moral crumple zones.”²

Further complicating these dynamics is the fact that not all players in the ecosystem assesses innovation the same way. A conservative global law firm evaluates change differently than a fast-growing startup. A Fortune 500 legal department operates under different pressures than a well funded scale-up. A west coast firm thinks differently than an east coast firm.

Distribution across the spectrum of legal firm types requires far more than mere product capability. It requires proximity to how each type of client thinks — about risk, about timing, about optics, about internal buy-in. The closer you are to those conversations, the more effective you can be at introducing change.

The upshot of all this: Innovative firms that distribute effectively tend to reflect the personalities & beliefs of the markets they serve. They aren’t built from a single archetype. They bring together operators with different professional backgrounds, different styles, different instincts about risk and speed. That internal range creates a closer read on the client’s reality. It makes it easier to introduce innovation gradually, without forcing the customer/client to absorb an immense amount of uncertainty all at once.

In institutional markets, technology doesn’t spread because it is powerful. It spreads because someone trusted made it feel safe.³

Conclusion

If legal work can be decomposed, accelerated, and systematized at scale, then the architecture of the industry will inevitably shift in some form. The question is not whether the mechanics will improve. They already have.

The more interesting question is how that improvement travels through institutions that are, by design, cautious.

Legal departments and law firms are not blank slates. They are layered systems of incentives, history, relationships, and risk management. When change happens, it tends to move through those layers unevenly: Expanding in some corners, stalling in others, accelerating where trust is already deep.

Maybe the next phase won’t be about who builds the most powerful models, but who earns the right to deploy them in increasingly consequential contexts. In high-risk markets, innovation doesn’t just need to work. It needs to feel safe.

And safety, unlike speed, rarely scales overnight.

1. The distinction between ordinary & durable distribution is worth emphasizing here. Some of the hottest new AI players are hitting world-class revenue growth metrics, which suggests that their distribution is highly effective. However, those metrics were designed in a world where software was sticky. Getting to $100M+ ARR as an example, suggested inevitability because switching costs were so high.

That’s not the world we live in any more. Switching costs have plummeted. Unless the software (AI or otherwise) is deeply embedded in workflows, via integrations, daily usage patterns, and customer mindshare—it is highly likely that these customers will churn at the end of the contract. Getting to $100M+ doesn’t signal inevitability any more.

This is especially true of AI that law firms buy for signaling purposes—ie. to show their clients that they’re being innovative. It is not enough for an AI startup to hit the “old world” milestones to reach inevitable “winner” status—the number today is likely far higher than that. Which will require strong net dollar & gross retention—both of which need high & growing usage.

2. A moral crumple zone is a concept that describes how responsibility gets displaced onto individuals when complex systems fail. It refers to situations where humans become the “shock absorbers” for institutional or technological breakdowns — much like the crumple zone in a car absorbs impact during a crash. When something goes wrong in a high-stakes system (like an AI-driven process), blame often lands on the human operator at the edge of the system, even if the root causes lie in design choices, incentives, or structural constraints.

3. This is something I have been thinking deeply about for many years, and will be the subject of a future article. It’s not a coincidence that history’s most effective legal tech distribution channels, e.g. Thomson Reuters & Lexis, invested huge amounts of resources to build trust with their customer base. They hired ex-lawyers as sales reps and account managers, and paid student ambassadors to represent their brand while future customers were still in law school. It’s why Biglaw business development still largely works in an old-school way, through quiet social proof & personal and professional networks. It all looks very clubby from the outside, but it’s effective precisely because relationships and reputation ensure accountability.