When AI can draft anything, reputation is the last moat left

There is a question I keep putting to the law-firm leaders I meet, and it makes most of them shift in their seats: when a competent AI can produce a serviceable contract, a passable letter of advice or a first-draft opinion in seconds, what exactly is your firm selling? The honest ones admit they are not entirely sure any more. For a long time the answer was implicit, you were selling the drafting, the knowledge, the sheer capacity to produce the document, and that answer is quietly being commoditised out from under the profession. The mechanics of legal work are becoming cheap. What is left, and what will only grow in value, is something the machines cannot touch: reputation.

I want to be precise about what I mean by reputation, because the profession tends to hear the word and think of Legal 500 rankings and a nice logo. That is branding, and branding is downstream of the real thing. Reputation is the accumulated answer to a single question in the client’s mind, can I trust this named human, and this firm, with something that genuinely matters to me? That question has never been answerable by a document, however well drafted, and it is even less answerable now that flawless-looking documents are effectively free. Trust is the product. Everything else is increasingly a commodity wrapped around it.

Consider what actually happens when the drafting becomes trivial. If a business can generate a plausible commercial agreement in an afternoon with a machine, the value migrates entirely to the moment of judgement, is this the right deal, have we seen the risk that is not on the page, whose name is on the advice if it goes wrong? None of those are drafting questions. They are trust questions, and trust does not scale the way software does. It is built slowly, in public and in private, through a track record that a competitor cannot download, copy or undercut on price. In a world of infinite cheap output, scarce trusted judgement becomes the premium good.

This is why I think individual named authority is about to matter far more than firm brand, which cuts against the whole partnership instinct to subsume the individual into the institution. Clients do not trust “the firm” in the abstract; they trust the partner they have seen think clearly in public, the specialist whose article actually helped them, the named human whose judgement they have watched hold up under pressure. As AI flattens the commodity layer, the premium attaches to identifiable people with demonstrable, well-evidenced expertise, the lawyer who is unmistakably the authority on a narrow, valuable thing. A firm’s real asset base is quietly shifting from its systems and its precedents to the credibility of the people whose names go on the work.

There is a paradox here that the profession would do well to sit with. The very technology that commoditises the routine work makes the authentic, human, verifiable signal more valuable, not less, precisely because it becomes harder to trust anything at face value. When anyone can generate a confident-sounding legal explainer, a confident-sounding legal explainer is worth almost nothing. What is worth a great deal is the confident explainer attached to a real, accountable, checkable human who stakes their reputation on it and can be held to it. Scarcity has moved. It used to sit in the knowledge; now it sits in the accountability.

The strategic implication is that firms should be investing in reputation as seriously as they invest in technology, and most are doing precisely the opposite, pouring money into AI tooling while leaving the credibility of their people entirely to chance. That is backwards. The tooling is a fast-depreciating asset that your competitors will have too by next quarter. The trusted authority of your named partners is an appreciating one that compounds over years and cannot be bought off the shelf. If I were running a practice today, I would be as deliberate about building the public reputation of my best people as I am about their utilisation, because in five years that reputation will be the thing clients are actually paying the premium for.

The uncomfortable corollary is that reputation, unlike a document, cannot be faked at scale for very long, and the same AI wave that commoditises drafting is also making audiences more sceptical of manufactured authority. A firm cannot prompt its way to being trusted. It has to earn the thing the slow way, through visible, consistent, genuinely useful expertise delivered by real people who put their names to it and are right often enough to be believed. That is inconvenient for anyone hoping to shortcut it, and it is enormously reassuring for any firm willing to do the patient work, because it means the moat, once dug, is very hard for a cheaper rival to cross.

So I would stop worrying about whether AI can draft your documents. It can, or soon will, and so can your competitors’ machines, and the client’s machine too. The question that determines who thrives is the older, more human one that no model has ever answered: whose judgement do you trust when it genuinely matters? The firms and the individuals who have spent years building a defensible, evidenced, named answer to that question are about to discover it was the only moat worth digging all along.