There is an important difference between using a tool and owning a capability. Most law firms that describe themselves as AI-forward are doing the former and calling it the latter. The distinction matters more than most firms have stopped to consider.
Using a tool means subscribing to a platform, configuring it for your workflow, and building your practice around it. Owning a capability means that the methodology, the institutional knowledge, and the competitive advantage are yours — regardless of which underlying technology you happen to be running at any given moment. One of those things can be repriced or discontinued by someone else. The other cannot.
Talking Heads built a whole song out of exactly this disorientation. In “Once in a Lifetime,” a man takes stock of a life he assembled one sensible choice at a time — the house, the car, the career — and is gripped by the sudden suspicion that none of it is really his. He keeps asking how he wound up here, how the machinery around him even works, whether the lovely house he is standing in was ever actually his to stand in. For a firm that has quietly threaded its practice through a platform another company owns, prices, and controls, that is not a passing mood. It is a balance-sheet question — and a quieter version of it is worth asking now, rather than on the morning the answer arrives uninvited.
In the previous post in this series, we looked at the operational risk of AI dependency: what happens when the tool goes down. This post is about the strategic risk: what happens when the tool that was your competitive advantage is no longer yours to rely on.
Same As It Ever Was
The legal profession has been through versions of this before, and the lessons have not fully taken.
Legal research was transformed by Westlaw and LexisNexis. Firms that made those platforms central to their practice benefited enormously — until pricing negotiations became a significant line item, and until the platforms themselves began differentiating their offerings in ways that created meaningful capability gaps between firms that could afford premium access and firms that could not. The tool that everyone used became the tool that stratified the market.
Document management, e-discovery, and practice management platforms followed similar trajectories. Startups built loyal user bases, got acquired by larger players, changed their terms, repriced their products, or quietly discontinued the features that had made them valuable. Firms that had built their workflows around specific capabilities found those capabilities altered or removed without meaningful notice.
The legal AI market is moving faster than any of those prior waves and consolidating faster still. The number of legal AI startups that will survive independently over the next five years is a fraction of the number that exist today. Acquisitions are accelerating. The terms that govern those tools today are not the terms that will govern them after the next funding round, the next acquisition, or the next pricing review.
That pattern is no longer a forecast. It is this quarter’s headlines. Legal-tech investment reached roughly $2.3 billion across more than a hundred deals in the first quarter of 2026, but a single trio — Relativity, Harvey, and Legora — absorbed close to two-thirds of it. Harvey closed a round in March valuing it at about $11 billion, only months after an $8 billion mark the prior December. Legora, its fastest-moving rival, pushed past a $5.5 billion valuation and went shopping, acquiring three companies in roughly three months. Thomson Reuters absorbed the legal-AI startup Noetica in February; Clio, fresh off a $500 million raise, bought the research platform vLex for about $1 billion. Bankers who track the space describe something on the order of a 200 percent jump in legal-tech deal activity year over year. Capital is concentrating around a handful of platforms at precisely the speed that makes the smaller tools — the ones a firm may have built a workflow around — the likeliest to be bought, folded in, repriced, or quietly shelved.
Living in a Shotgun Shack
When a law firm subscribes to an AI platform — even a sophisticated one, even one they have configured extensively — they are renting. The underlying model is not theirs. The training data is not theirs. The infrastructure is not theirs. In most cases, the workflows they have built on the platform are not portable to a different platform without significant rebuilding.
That is not necessarily a disqualifying problem. Most technology is rented in this sense, and renting has real advantages: someone else maintains the infrastructure, absorbs the development cost, and handles the updates. The question is not whether you are renting, but whether you understand what you are renting and what you would do if the rental terms changed.
The firm that has built a competitive advantage on a rented platform has built something real — but something fragile. The advantage lasts exactly as long as the platform’s terms support it. When a competitor can subscribe to the same platform and access the same capabilities, the advantage narrows to configuration and execution. When the platform changes its terms, raises its prices, or gets acquired by a competitor’s preferred vendor, the advantage can disappear entirely.
And the rent does not hold still. Across 2025 and into 2026, the major AI providers began moving enterprise customers off the flat-rate plans that made budgeting predictable and onto per-token, usage-based pricing that charges for what agentic workflows actually consume — which, at current adoption levels, has proven to be far more than the early spreadsheets assumed. One large technology company reported burning through its entire annual AI budget in four months. Providers marching toward public markets have every reason to show margins, and the direction of pricing is not down. A firm that built its efficiency model on the introductory rate is, to borrow the song’s image, drifting toward the moment once the money runs out — still standing in the same practice, suddenly unsure it can afford the house.
How Do I Work This?
Owning your methodology does not require building every tool from scratch. It requires ensuring that your competitive advantage is embedded in something the other side cannot simply purchase.
In the framework we described in our field guide to AI in legal practice, the Expert level is defined precisely by this distinction. Expert-level AI use involves proprietary tools — purpose-built systems that embed institutional knowledge, practice-specific judgment, and workflow design that took significant time and expertise to develop. The gap between knowing what those tools do and being able to replicate them quickly is real, and it does not close just because a competitor understands the general approach.
At lower levels, the emphasis shifts from proprietary tools to proprietary process. A firm that has developed clear, documented AI workflows — specific sequences of direction, review, verification, and quality control that reflect how that firm practices — owns something that a competitor using the same platform does not automatically have. The process is the differentiator, not the subscription.
This distinction also matters for client conversations. When a client asks what makes your AI use different from another firm’s, the answer should not be the name of the platform you subscribe to. The platform is available to anyone. The methodology — the specific way your firm directs AI, reviews its output, integrates it into legal judgment, and verifies its accuracy — is what you have actually built. That is what is worth describing, and protecting.
Letting the Days Go By
Every firm that relies on AI should have a clear-eyed view of its platform exposure: which tools are critical to which workflows, what the switching cost would be if those tools changed significantly, and what portion of its claimed AI advantage would survive a major platform disruption.
That analysis will look different at different firms and different levels of AI sophistication. For a Foundation-level practice, the switching cost is relatively low — general-purpose AI tools are largely interchangeable. For a firm that has built Advanced or Expert-level workflows on a specific platform, the switching cost may be substantial, and the strategic question is whether that investment is protected by something the firm actually owns or simply by the platform’s current terms.
The firms that are building most durably are the ones that treat platform selection as a long-term strategic decision rather than a procurement choice. They are asking not just which tool is best today, but which tool gives them the most control over their own capability going forward — and where they need to own the methodology entirely, independent of any vendor’s decisions.
You May Ask Yourself “Well, How Did I Get Here?”
Not every firm can build proprietary AI tools, and not every practice area requires that level of investment. But every firm can take steps to reduce platform dependency: documenting workflows so they are not locked in a single system, diversifying across multiple tools rather than concentrating on one, and ensuring that the institutional knowledge behind AI workflows is held by people, not just by the platform.
The competitive advantage worth having is one that belongs to you. A subscription that anyone can buy, at terms that someone else controls, is a starting point — not a destination. It is a beautiful house with someone else’s name on the lease — and, same as it always was, the lease is theirs to rewrite.
In the next post in this series, we look at a different dimension of AI practice management: the supervision obligation. When AI is doing substantive legal work, who is responsible for what it gets wrong — and what does real supervision actually require?
If questions about AI strategy and practice management are relevant to your firm or your clients, we are glad to have that conversation.
