AI may mean death of legal apprenticeship
Work now being dealt with by agents is what helped trainees learn, says Rob Aberdein
Every senior lawyer I know learned the job the same way: slowly, repetitively and, if we are honest, partly at the client's expense.
We read thousands of documents. We drafted letters that came back covered in red ink. It was rarely glamorous, but it worked. Somewhere in all that mechanical effort, judgment was quietly being built.
That apprenticeship is about to disappear.
I have argued AI will change how firms bill and how legal work gets done. Let's assume I'm right, and this future arrives faster than most expect. If agents take on reviewing, researching and drafting, there is a consequence we have barely discussed: the work they absorb first is the work we have always given to trainees.
These tasks were never just cheap labour. They were the curriculum. A junior lawyer who has read 500 leases knows what an unusual one looks like. One who has drafted the same letter 20 times starts to understand why the partner changes it every time.
If an agent now does that work in minutes, the efficiency gain is real. So is the risk. We could build firms that are faster and more profitable today while hollowing out the pipeline of lawyers capable of supervising those agents tomorrow.
Here lies the paradox of the agentic law firm: the agent performs the task, but the lawyer remains accountable.
But accountability requires competence. A lawyer checking an AI-generated contract needs to know what a good one looks like, where the traps sit and when something that reads perfectly is quietly wrong. That instinct doesn’t arrive with a practising certificate. It is earned through repetition, error and correction.
Remove the repetition and we risk producing lawyers who can operate the tools but can’t tell when they’re wrong. That is not a technology problem. It’s a professional standards problem.
Training by design, not by osmosis, is a question vexing me. If, as an unapologetic advocate for this brave new world, I urge firms to hand junior work to agents, I owe an answer: how do we take young lawyers with us and ensure the next generation is properly trained and developed?
I don't have it fully solved, but some things are clear. The answer is not to slow adoption or invent busywork; clients won't pay for it. It is to stop relying on osmosis and start designing how lawyers learn.
That means getting junior lawyers closer to clients sooner: on the calls, preparing the advice, watching negotiations live. It means using AI as a sparring partner rather than a substitute, asking trainees to critique an agent's draft and find its errors. It means partners treating feedback as core work - and tracking what each junior has actually been exposed to.
None of this is complicated. It is simply good supervision.
For firms that get this right, there is a commercial prize. Graduates weighing up law increasingly ask what they will actually learn, and how quickly. A firm that can answer with a structured, AI-literate training model will attract better people. A firm that answers "you'll pick it up as you go" will lose them.
The profession has spent years asking what AI can do for lawyers. The harder question is what lawyers will need to be able to do once AI is doing the rest.
Get that wrong, and we will have built the most efficient law firms in our history, with nobody ready to lead them.
Rob Aberdein is General Counsel and Group Director, Legal & Strategic Growth, Hoxton Wealth, and Managing Director, Simpson & Marwick Group.