A first-year associate now spends less time than their predecessors putting together the first draft of a due diligence memo. The relevant clauses have already been extracted, the outliers marked, and something that looks, at first glance, like competent legal writing has been produced. But what kick-starts her career is what she does with that draft. One reason a legal transformation consultancy like Everingham Legal spends so much of its time advising legal departments on how to adapt, rather than just which software to buy, is that firms and in-house teams are still working out what this shift means for hiring, training and promotion.
The Draft Is Less Expensive, the Judgment Is Not
Drafters have been a substitute for competence for most of the life of the profession. Junior lawyers made their names and their instincts writing first drafts under pressure and having them marked up by someone more senior. The apprenticeship model assumed that writing was the hard part. Increasingly it isn’t. The training ground for junior lawyers is shifting beneath their feet as machines can now attempt contract review, legal research summaries and first-pass drafting in seconds.
That does not mean junior lawyers are less needed. It changes what they need to be good at earlier on in their careers. What they have to learn is judgment, but they have to learn judgment by judging volume, not by producing volume. That is a different skill, and a much, frankly, much harder skill to teach.
Reading Machine Print the Way You Would Read a Junior’s Work
The most pressing skills gap is in the way lawyers treat AI-generated material. Too many consider a polished looking output a completed one, mainly because it is free of the hesitations or hedges a junior colleague might insert. That confidence is a facade. A tool can miss a jurisdictional nuance, misstate the current status of a regulation, or apply a template clause to a fact pattern it does not in fact fit. And it can do all of these things while sounding entirely sure of itself.
The lawyers who use these tools well use them the way a good supervising partner uses a junior associate’s memo: useful, often accurate, but never beyond scrutiny. That means checking citations against primary sources, sense-checking conclusions against known facts of the matter, and being willing to throw away an output wholesale when it doesn’t stand up. Critical evaluation is quickly becoming as fundamental a legal skill as legal writing itself.
Prompting Is an Art, Not a Shortcut
There is a temptation to see prompting as a technical add-on. Something for the IT team or a keen junior to sort out. In practice, you need something closer to the instinct of legal drafting to get useful output from a legal AI tool: knowing which facts matter, what ambiguities need to be resolved before you can trust the tool, and what the tool is structurally incapable of doing.
Understanding the limitations of a tool is just as important as understanding what it can do. A system trained primarily on public case law will approach a simple contract review differently than a matter with a strange regulatory overlay or a novel commercial structure. The lawyers who know where the seams are, where the training of the tool gets thin, or where the assumptions are flawed, get far more reliable results than those who treat every output as equally trustworthy.
Risk Assessment Center
In a world where routine drafting is automated, the value a lawyer brings is further up the chain, in risk assessment and strategic judgment. Clients were never paying for the actual drafting of a clause. They were paying for someone to tell them if that clause protected them, how much it would cost them if it didn’t, and how a counterparty or a court might read it differently than they intended.
That kind of judgment doesn’t come from a template. It’s based on pattern recognition, learned over years of watching deals go sour, disputes get ugly and clients get burned by language that looks fine on paper. If lawyers want to stay relevant, they must intentionally develop this kind of judgment instead of assuming it will develop on its own, because the repetitive drafting work that used to develop it by accident is disappearing.
Comfy Working With Non-Lawyers
Legal departments are increasingly employing data specialists, legal technologists and process designers – people whose job it is to make the tools work well, not to practice law. Lawyers who can work with this group, who can explain what a clause needs to accomplish, who can point out where a workflow does not match how matters actually unfold, tend to get far more out of the technology than those who treat it as someone else’s problem.
It’s less about learning how to code and more about developing enough technical literacy to have a useful conversation. Like how a good litigator doesn’t need to be a forensic accountant, but needs to know enough to ask the accountant the right questions.
What artificial intelligence is unable to do yet
None of this replaces the aspects of legal work that were never really about creating text. Negotiation still requires reading a room, feeling when a counterparty is bluffing, and knowing when to hold a position and when to fold. Client relationships are still built on trust earned over years, often in moments that had nothing to do with the substance of a matter. Ethical judgment, weighing competing duties, knowing when to tell a client something they don’t want to hear, is stubbornly human territory.
Everingham Legal’s work with law firms and in-house teams tends to come back to this same point: the technology is changing the way legal work gets produced, but it hasn’t changed what clients are ultimately paying for. The lawyers who will build careers that will last as the profession continues to evolve are those who invest in critical evaluation, technical fluency and sharper judgment, while doubling down on the negotiation and relationship skills that no tool can replicate.
