“[Productivity] gains only come when companies use AI to redesign processes and ultimately rethink whole business domains. That’s where the step-change in efficiency and growth will come from. To get there, the foundations must be right — clean, well-governed data; secure and interoperable systems; and people who understand how to work alongside AI.”
This observation from Jonathan Keane, Strategy and Consulting Lead at Accenture for UK, Ireland and Africa, was a highlight of the recent Financial Times Special Report on AI. It was meant to apply to a range of businesses, but I think it lands most impactfully in the legal services space.
In many respects, law firms are ideal environments for AI to take root and flourish. Generative AI can draft lawyer-like or near-lawyer-like content in a matter of seconds — emails, agreements, factums, revisions, and legal opinions, to name just a few. When grounded in reliable legal data and reinforced with controls like retrieval-augmented generation and citation-checking, AI’s legal output becomes markedly more dependable. It almost seems custom-designed for law firms and their knowledge-work focus.
And increasingly, document generation is looking like AI’s opening act. Agentic AI, as this excellent primer by LegalTechHub explains, consists of systems that go beyond mere output generation to plan and take actions toward a goal, iterating across steps rather than waiting for a human to drive each interaction. They are “software components designed to operate with delegated authority within defined bounds.” When fully mastered (and duly guardrailed), agentic AI could potentially bring law firm productivity to even greater heights (although not, within the near future, to a point where lawyer oversight is unnecessary).
But therein lies the problem, and the key to understanding how, in other respects, AI is a terrible fit for law firms — because most law firms still define “productivity”in 19th-century terms.
As I’ve explained elsewhere, law firms measure inventory in hours and understand “productivity” in terms of volume: The more hours billed and collected, the more money generated — and the more of those hours that come from leveraged assets like associates and non-equity partners, the more profit produced. AI’s capacity to generate lawyer-like output is remarkable; but the speed with which it works is what’s really game-changing, because it renders “time worked” virtually irrelevant as a measure of value.
But the disconnect goes deeper than that. The growing capacity of AI to carry out work — not just one-off requests through a chatbot interface, but also entire chains of activity through the use of AI agents — means that AI is displacing lawyers as the primary means by which legal task are performed. We like to say that “AI augments the lawyer,” but I suspect that within a few years, especially in high-volume areas, we’ll find ourselves saying that “the lawyer augments the AI.”
In other words, we are rapidly heading towards a role reversal in legal service production. The foundational unit of productivity in law firms has always been “the lawyer,” through whom all activities of value flow and around whom staff and systems are assembled and arranged. Sooner than we think, AI will displace lawyers at the gravitational center of legal productivity: The machine will simply be more productive and ultimately more effective at coordinating and carrying out legal tasks than lawyers are.
Notice that I’m saying displacement, not replacement. Lawyers aren’t going to disappear (or they shouldn’t, unless they really play their cards badly); but they are going to switch seats. They will allow AI to be installed and operate as the productivity engine of legal work, and will dedicate themselves to those legal service activities for which people are better suited than machines: building human relationships, expressing human empathy, performing human advocacy, and exercising human judgment.
That’s the post-AI future I see for the legal profession. But I’m increasingly of the view that that future cannot be realized within the structure of traditional law firms. Law firms, as we’ve always known them, are lawyer-centric entities in every respect: Everything of value and importance, above all the delivery of legal services, flows through the lawyer.
AI simply isn’t compatible with that kind of model. In fact, I believe that AI is fundamentally antagonistic towards the traditional law firm. For AI to be truly impactful in the legal services context, a new model becomes necessary, one that incorporates and integrates AI as the default option for production, and eventually as the fundamental engine of productivity.
This is why I’m keeping a close eye on “AI-native” law firms and “AI-first” law firms — legal business entities built around AI engines rather than lawyer engines. To be clear, these entities are still in their infancy, and a lot of things need to go right, primarily with how they manage the economics of output verification, for them to really take off. But I’m coming to think these kinds of firms do hold the key to the future of legal work. The more deeply traditional firms try to use AI, the more frustrated they may become, because they’re deploying a technology that’s incompatible with the way they measure value and organize production.
Photographic film companies couldn’t survive the invention of digital photography, and the commissioned-broker model was swept away by electronic trading. Every so often, a new technology comes along that resets all the mechanisms and changes all the assumptions about how work gets done in an industry.
I think lawyer-centric law firms are poised to join them. I think we need to get ready for AI-centric law firms.
The post Towards AI-Centric Law Firms appeared first on Slaw.