Posted by Jack Shepherd, senior principal business consultant at Legal Futures Associate iManage
Shepherd: Low-value tasks are the ones AI can solve today
The legal industry is excited about AI. That’s good. But the direction of that excitement isn’t always useful.
A lot of attention is being directed towards AI tools that aim to replace or support high-value work: drafting contracts, writing legal advice, analysing case law. These are ambitious projects, and they’re intellectually satisfying to work on. But many of them are incredibly hard to get right.
And more importantly, they’re not necessarily what lawyers need most right now.
There’s a category of legal work that’s boring, repetitive and still done manually. Think fixing formatting in Word, rebuilding tables, changing first person to third, removing comments, cleaning redlines.
Think document bundling, renaming PDFs, renumbering exhibits. Think time recording, email filing, setting up matter folders.
Nobody’s excited about this stuff, but it still eats up hours of a lawyer’s week. These things are often the reason why lawyers are still in the office at 2am. And clients don’t pay for it.
It’s these tasks – the really dull ones – where AI could make a visible difference quickly. They’re low-risk, mostly standardised, and don’t involve any complicated legal reasoning. If a model makes a mistake while tidying up a formatting issue, the consequences are minimal. If it hallucinates a new clause in a contract, the situation is very different.
But too often, legal tech vendors chase the high-value processes instead. They market tools that generate legal opinions or write entire memos. This sounds impressive, but it assumes that these are the tasks lawyers want to offload. Often, they’re not.
A lot of lawyers actually like drafting. They find it valuable, even enjoyable. They use it to clarify their thinking. What they don’t like is spending half an hour manually accepting track changes across 20 documents before a deal closes.
There’s a gap between what lawyers enjoy and what they find frustrating. Legal tech should start by removing the things people find frustrating. That means focusing less on headline-grabbing use cases and more on the tedious admin that makes lawyers groan.
These are the jobs that fall through the cracks because nobody wants to think about them. Many don’t think about them because many don’t know about them (because they haven’t been lawyers themselves, or haven’t spent enough time speaking to lawyers). But they’re also the jobs that waste the most time across the profession.
There’s another angle here: context. Not all drafting is equal. Writing a bespoke clause to deal with a tricky risk allocation in a cross-border finance deal is a valuable exercise. Copying and pasting boilerplate text about third-party rights is not.
The same goes for legal research. Crafting a novel argument from obscure case law is valuable. Writing up standard advice on directors’ duties is more run-of-the-mill.
Even within high-value processes like drafting or research, there are low-value tasks worth automating.
But you need to be clear about what those are. Be specific. Don’t just say your AI helps with contract drafting. Focus on the specific task within the contract drafting process. Say you resolve pain points around replacing placeholder terms, updating definitions, and checking clause references. That’s where the pain is.
There’s also the question of what’s lost when you automate. Some manual tasks create value by accident. Take court bundling. It’s boring, it’s slow, and it often falls to the most junior team members. But it also forces someone to become intimately familiar with the documents.
That knowledge can become useful in hearings or strategy meetings. Automate the bundling and you might lose that familiarity.
So if you’re going to automate the work, think about how you’ll replace the value it created. Sometimes, losing that value is worth it in face of the efficiencies – sometimes it is not.
This doesn’t mean we avoid automation. It means we do it thoughtfully. If the process has no intrinsic value, it should be automated. If it does have value, find a better way to achieve that value while still getting rid of the tedious work.
AI should help lawyers move up the value chain, not just press buttons faster.
One reason we don’t see more focus on low-value tasks is that they’re not very glamorous. Solving document formatting doesn’t make headlines. But these are the fixes that free up time and improve morale.
More importantly, they’re the kinds of problems AI can actually solve today. The use cases are narrow, the data is simple, and the risk is low.
If we want AI adoption in legal to grow, we need to start with the things people already want to get rid of. Not the high-stakes work they enjoy doing, but the repetitive tasks they resent. That’s where trust gets built. That’s where lawyers start to believe AI is useful.
Of course, it’s fine to pursue the more complex problems too. But the balance is off. Right now, too much of the conversation is about replacing the clever stuff. We need more focus on replacing the tedious stuff. Automate the keyboard, not the brain.
Leave a Comment