The technological lawyer in the age of AI

Artificial intelligence (AI) is no longer merely a tool lawyers use. It is becoming an environment in which legal method may be organised, repeated and scaled, says Dr Corsino San Miguel. The age of AI will not only change the desk. It will demand a more technological lawyer.

From default AI to the Claude-native moment

In early February, I suggested in these pages that legal AI would not arrive as a prophecy suddenly fulfilled, but more quietly, by settling into the surfaces through which lawyers already work. It would become default not because the profession had resolved all its doubts, but because the desk itself would change shape.

At the end of February, Zack Shapiro’s now-viral post on X, The Claude-Native Law Firm, seemed to give that argument a face. Not simply because the post drew extraordinary attention by legal standards, with coverage reporting almost eight million views. It mattered because it brought into view something many lawyers had only vaguely sensed: that AI is no longer just a tool we consult, but may be becoming an environment in which legal work is increasingly carried out.

What Shapiro appears to have done is more interesting than simply using Claude to draft. According to the reporting, he built Claude into the working method of his two-person boutique, Rains LLP. Rather than treating the model as a blank assistant each time, he used custom skills and agentic workflows so that Claude could work in a way more closely aligned with his preferred structures, review habits, style and judgement about what should be flagged, escalated or rewritten.

That is why the moment matters.

It is not because one American lawyer found a clever way to use Claude. Clever lawyers have always found clever ways to work. Nor is it because legal AI has somehow solved law. It has not. The significance lies elsewhere. What Shapiro appears to describe is a way of working in which AI is not used simply to answer isolated questions or produce first drafts, but to carry part of the lawyer’s own method.

An earlier thread in the argument

That possibility has been quietly building for some time. Three years ago, I wrote in this Journal that AI might offer smaller Scottish firms and sole practitioners a genuine opportunity to narrow the gap with larger competitors. My argument then was largely about market structure. Smaller firms, I suggested, could use AI to improve efficiency and sharpen capability without having to reproduce the full cost base of larger practices, while medium-sized independents risked being squeezed between agile boutiques and resource-rich larger players.

The recent Claude-native discussion does not overturn that argument. It sharpens it.

What once looked like an opportunity for smaller firms to catch up in efficiency may now be developing into something more interesting: an opportunity, in some areas, to compete through encoded method rather than headcount alone.

When legal method becomes explicit

That phrase matters. Much of the conversation around legal AI still proceeds as though the main question were which product is best, which wrapper is safest or which interface is most polished. But the more interesting question may be different. It may be whether the real competitive asset lies less in the tool itself than in the lawyer who can describe, with enough clarity and discipline, how the work should be done.

The scarce asset, in other words, may not be the software alone. It may be the lawyer who can make method explicit.

This helps explain why the Shapiro episode landed with such force. His reported use of Claude was not presented as casual prompting or mere convenience. It was presented as a way of making his own professional method more reusable.

Lawyers have long relied on precedent banks, standard forms, house styles and internal know-how. But much of the real craft of practice has always lived somewhere more elusive: in the margin notes, the instincts, the way a particular lawyer frames a risk, structures a clause, narrows an issue, softens a line or knows when a point must be escalated rather than explained away.

Traditionally, those things travelled slowly. They moved through apprenticeship, supervision and repeated exposure. A junior lawyer learned by working near someone more experienced, by absorbing not only what was written, but how and why it was written that way.

What makes the current moment different is the suggestion that some of that craft can now be rendered into instructions, routines and repeatable workflows. Not the whole of legal judgement, certainly. But enough to matter.

Why this changes the position of smaller firms

For smaller firms and boutiques, that matters a great deal. The old relationship between size and capability may not disappear, but it may loosen. A small practice with a strong method and the imagination to encode it may acquire leverage once associated more naturally with institutional scale. Not the whole of scale, of course. Larger firms will still have deeper benches, broader client relationships, stronger governance functions and, in many cases, better access to authoritative legal infrastructure.

But the boutique may gain something newly potent: the ability to operationalise its distinctiveness.

That, I think, is the real continuation of the February argument. If legal AI becomes default by embedding itself into the work surface, then the next question is what else becomes default with it. The answer may be: more of the lawyer than we first supposed.

Not only our drafts, but our thresholds.
Not only our templates, but our habits.
Not only our documents, but our preferred routes through a problem.

Why lawyers may be more prepared than they think

And that is where the strategic issue begins.

The question is no longer simply whether firms should ‘use AI’. That formulation is already too shallow. The more serious question is how lawyers should approach a technology that responds, above all, to clear instruction, disciplined framing and structured delegation. In that respect, lawyers may be better placed than they sometimes assume.

After all, much of legal practice already involves a form of prompting. Partners brief associates. Senior lawyers structure a problem, identify what matters, define the scope of the task, specify the form of the output and indicate what should be flagged urgently or treated with caution. Good legal supervision has always depended on the quality of the instruction. What AI does is make that habit newly visible.

The lawyer who knows how to frame a task well, sequence reasoning clearly and communicate expectations precisely is not learning an entirely foreign art. They are extending a familiar one into a new medium.

Why lawyers may need to become more technological

But that is only the beginning. The technological lawyer is not simply the lawyer who prompts well. It is the lawyer who begins to think more structurally about how legal work is organised, repeated and improved through technological means.

That requires something more than curiosity about the latest tool. It requires a shift in professional posture. The technological lawyer asks not only, “Can this system help me with this task?” but also, “What part of my method can be made clearer, more portable, more consistent, more scalable?”

In that sense, Shapiro matters not because he offers a template to be copied wholesale, but because he acts as a kind of probe. He tests, in unusually public form, what happens when a lawyer stops treating AI as an occasional assistant and starts treating it as part of the architecture of practice.

That is the deeper challenge now facing the profession. Lawyers will need to become more technological not in the shallow sense of becoming amateur engineers, but in the professional sense of learning to think in systems, workflows, interfaces and operational design. They will need to understand not only legal reasoning, but how legal reasoning is increasingly expressed through technological environments.

That does not remove the risks. Firms still need to think carefully about authority, confidentiality, governance and the danger of allowing valuable legal method to become too easily absorbed into external platforms. High-stakes legal work still requires trusted sources, editorial control and human judgement. But if the profession limits its response to caution alone, it risks misunderstanding the moment. The issue is not only defensive. It is also formative. A new kind of legal competence is beginning to emerge: the capacity to render good judgement usable within technological systems without surrendering judgement to the system.

The desk, and the lawyer

But the Shapiro moment suggests that lawyers are not standing before an alien machine with no native advantage. On the contrary, one of the profession’s oldest strengths – the ability to direct complex reasoning through language – may turn out to be one of its strongest assets in the age of AI.

The technological lawyer, then, is not the lawyer who abandons craft for automation. It is the lawyer who learns how to carry craft into a new medium.

The old desk held books, papers, precedent banks and the visible traces of a lawyer’s trade. The new desk may begin to hold something more intimate: a working model of how the lawyer thinks.

That is why the Shapiro moment deserves attention. Not because it is a prophecy. But because it is a proposal, newly visible, for where legal value may start to reside.