TL;DR
- The overnight data room grind was never a training strategy. It was thought of as the only way to get the work done.
- Judgment doesn't come from reading 400 contracts. It comes from reading them alongside the reasoning for why a clause matters.
- The junior's job is now harder, not easier: interrogating analysis requires understanding it in a way that producing a list never did.
For decades, being junior in an M&A team meant one thing: find the needle in the haystack.
One big law partner told us that when he kicks off a deal, the first thing he does is put a junior lawyer on a data room overnight to pull every termination and change of control provision. That lawyer is looking for concepts they don't fully understand, with little direction, on client time, so a partner can red-line the result in the morning.
We've called that training for thirty years. It's neither an efficient nor effective way to get the work done.
The real question isn't whether AI takes junior lawyers' jobs. It's which parts of that job were ever worth keeping.
The objection
If juniors stop grinding through data rooms, where does judgment come from?
It's a fair question, and it assumes the grind was building judgment in the first place. Mostly it wasn't. Reading 400 contracts without knowing what you're looking for builds stamina, not pattern recognition. Reading 400 contracts alongside the reasoning for why a clause matters, how to apply this to the legal concepts that underpin it, that's what builds training and legal judgment.
What the job looks like now
A second-year still opens the contract. They still form their own view. What's changed is that they start from a first pass with evidence attached, so their attention goes to the three provisions that require their legal analysis, not the ninety-seven that don't. The job has shifted from producing the analysis to interrogating it. The validation layer.
That's a harder job, not an easier one. Validating someone else's reasoning requires you to understand the reasoning. But it means you can get up the curve quicker, in a more targeted manner.
Where we sit
Deeligence is built for that handover. Every flagged issue ties back to the source, with the reasoning available, so a junior always has something real to check against. We benchmark our outputs so quality doesn't depend on how well a lawyer happens to prompt, which is nearly impossible for firms to gauge on their own. Our legal engineers came from BigLaw and have built their teaching into the system, so the guidance comes with the output.
It's the equivalent of a senior lawyer sitting next to a junior on every document.
AI doesn't replace professional responsibility. It just means juniors spend their time exercising it instead of burning the night getting to the point where they can.
It's the difference between learning the job by getting burned and learning it by getting good. We think that's the future worth building.