For decades, private practice law was protected by scarcity. Talent from elite universities was limited. Human attention was finite. Even the best lawyers could only work so many hours. This natural constraint supported stable pricing and professional hierarchies. In Michael E. Porter’s Competitive Strategy, this absence of “excess capacity” was what kept industries structurally healthy. Legal, for a long time, benefited from exactly that.

AI has now removed this constraint. Systems can review, draft, research, and monitor at industrial scale. What once depended on scarce human intellect can now be replicated endlessly. The profession is experiencing something new: near-infinite intellectual capacity. As in travel, finance, and accounting before it, this excess capacity puts downward pressure on prices and makes mid-tier, undifferentiated work vulnerable to commoditisation.

At the same time, AI is following a familiar path: from supplier to competitor to platform. What began as a tool for lawyers, is increasingly serving clients directly. Next, it risks becoming the operating layer of legal work itself. If firms rely entirely on external systems, they risk becoming service providers inside someone else’s infrastructure. History suggests that incumbents who fail to internalise such shifts lose strategic control.

This shift also reflects a deeper change in what legal services mean. As Richard Susskind has argued, the profession is moving from “the ambulance at the bottom of the cliff” to “the fence at the top.” Instead of reacting to problems, AI systems will increasingly prevent them. Legal becomes embedded infrastructure: continuous, preventative, and data-driven. That is not the end of lawyers - but it is the end of purely reactive human led service model.

Not all competition created by AI is destructive. Porter distinguished between “bad” competition that destroys industry economics and “good” competition that strengthens them. Race-to-the-bottom automation is bad competition. But firms that use AI to deepen relationships, improve governance, and build better systems are engaging in good competition. The opportunity is not to fight technology, but to shape how it is integrated.

One promising direction firms could take is platforming without disintermediation. Instead of training their replacement and ultimately competing with their supply chain, firms can use technology to productise their own collective expertise, empowering their own clients. Models such as ours at Jylo illustrate this approach: no reuse of client data, no competitive exploitation, and support for on-premise, private cloud or sovereign deployments. Here, technology amplifies professional expertise for law firms rather than replacing it.

For private practice, this points to a constructive strategy. The most resilient firms will own workflows, institutional memory, and client-facing systems. They will become architects of legal capability, not just providers of hours. By co-building platforms with clients, protecting data sovereignty, and embedding AI into trusted relationships, firms can create defensible, differentiated positions. By contrast, low-barrier, out-of-the-box AI workflows merely dilute the exceptionalism that leading firms have built over decades, turning hard-won expertise into interchangeable output.

In an age of excess machine capacity, the scarcities that matter most are judgment, context, trust, and accountability. These cannot be automated away. AI raises their value. Firms that invest now in “good competitor infrastructure” - systems that strengthen professionals rather than commoditise them - will not only survive this transition. They will define what high-value private practice looks like in the next era.

And for everyone who once said, “Jylo is too advanced - our lawyers will never build those workflows,” the reality is now becoming clear. As AI reshapes the economics of legal services, firms no longer have the luxury of standing still. Building intelligent, client-focused systems inside platforms with aligned incentives - is no longer an optional experiment - it is becoming a prerequisite for remaining relevant in the next era of private practice.