ost people outside the law don't realise this but even at the very top of corporate and commercial law, roughly 80% of what a lawyer does is admin.
We call it The Admin Iceberg. The judgment work, the part clients actually want from a lawyer, sits above the waterline.
- 01
Above the waterline · 20%
The judgment work. The part clients actually want from a lawyer, and the part good lawyers actually enjoy.
- 02
Below the waterline · 80%
A huge mass of mechanical work that still has to be done and still has to be right, but requires almost no legal judgment at all.
Below the waterline
Underneath it is a huge mass of mechanical work that still has to be done, still has to be right, but requires almost no legal judgment at all. And in the traditional model, clients pay premium hourly rates for all of it.
Onboarding a client, KYC and compliance, Billing, writing routine emails, importing a document from an email, running redlines, building an issues list, reflecting agreed changes back into a draft, digging through old files to check what was done last time, doing the first pass of research, packaging it all up for a senior person to review. This work has to be right. But it's work that, honestly, a capable university student could be trained to do.
The other 20%
Then there's the other 20%. This is the part that actually needs a lawyer. How material is this risk, really? In the context of this deal and what's commercially sensible, should the client accept it, or is it the kind of risk that bites? If it bites, how do we mitigate it? What fallback positions can we live with? What's the pragmatic path through? It's understanding how a suite of documents fits together, how they sequence, where the interdependencies are. This is the intellectual core of lawyering, and it's the part good lawyers actually enjoy.
The traditional model charges you premium hourly rates for both. You pay the same senior lawyer's rate for the redline reconciliation as you do for the judgment call.
How an AI-native firm splits it
An AI-native firm treats the two sides completely differently.
On the 80%, we're ruthless. We use AI to completely eliminate it.
On the 20%, we don't try to automate the judgment, because you can't, and you shouldn't want to. Instead the lawyer works with an AI copilot that knows them, knows the client, knows the firm, and knows the research. You talk to it like a colleague, an intellectual sparring partner. It'll offer a view, but the lawyer is under no obligation to take it.
In fact we push our lawyers to argue with it, to interrogate it, to use it to instantly pull up what we agreed last time or how a similar point was handled on another matter.
The judgment stays human. The friction around it disappears.
The result
The result is straightforward. Lawyers are happier, because they spend their time on the interesting 20% instead of drowning in the 80%. The work is faster. And clients stop paying City rates for tasks that never needed a City lawyer in the first place.
Originally posted on LinkedIn.