Why AI hasn't disrupted in-house legal

In May 2023, Chegg’s CEO told investors that ChatGPT was hurting the company’s new customer growth. Its stock dropped 48%.

This makes sense. Chegg had a direct AI problem. Students paid Chegg for online learning, and ChatGPT could produce that for free.

We heard similar predictions about what would happen to in-house lawyers. But none of that has happened so far. In-house legal teams are adopting AI, but the change looks more like steady improvement rather than a function being overturned.

In this post I’m going to try to answer why that is.

Here’s my mental model: Disruption can play out at a company level or a team level.

At company level, a useful test is whether AI can materially change the economics of what customers are buying. Chegg is a perfect example of AI messing with a company’s main thing.

A logistics company like FedEx sits at the other end of the spectrum. Its main thing is delivering packages. I imagine AI can improve efficiency through better routing, less admin etc etc. But it doesn’t do the main job, which is the packages.

So Chegg faced a much greater disruption risk and had to rethink what customers were paying for. FedEx faces a more comfortable challenge: using AI to improve margins without changing the underlying job.

At team level, the amount of disruption to expect depends on how closely that team’s performance affects the main product of the company. If it is tightly linked (eg, developers at a software company), you can expect a lot of organisational pressure to see if AI can materially change the team’s output. After all, that output directly affects the company’s revenue potential.

But for a support function that’s much less likely to be true.

Let’s now look at how in-house legal teams fit into this mental model.

Median total in-house legal spend is roughly 0.5% of company revenue, according to 2024 benchmarking by the ACC and Major, Lindsey & Africa.

Imagine an enormously ambitious programme where AI somehow halves total legal spend. The company’s margin would increase by roughly 0.25 percentage points.

This is obviously a good result. But it is a different order of magnitude from halving the cost of producing or delivering the company’s core product. AI initiatives tied directly to that product can therefore support much larger ambitions than legal cost-saving alone.

Most legal teams will of course still be expected to make their budget go further by using AI. It just isn’t usually existential. (Could be different for law firms though!)

There is also a limit to how rational it is for in-house legal to aggressively chase efficiency, without treading quite carefully.

The company has a legal team to keep deals moving without accepting undue risk. Lowering legal cost is attractive, but not if it weakens the judgment and controls the function was hired to provide. There’d be no point in that.

As a result, full automation of the in-house legal team without checks and balances on quality offers an unattractive trade. The potential saving is a percentage of a relatively small cost line. But the downside is potentially worse performance on the dimension that justifies the team’s entire existence.

So in-house legal’s focus should be less on using AI to achieve internal cost savings, and more on improving the metrics that affect the core business. That’s a much bigger lever to creating value.

If eg an AI tool can help reduce ‘time-to-close’ from 8 weeks to 2 weeks, sales may close a deal this quarter rather than next. Or a supplier may start work before a critical deadline. Those gains won’t appear as lower legal spend, but the business will feel them.

But as we saw above, speed alone isn’t enough. Moving faster by missing important risks would be a poor trade.

So legal AI tools should be judged on whether they help the business move faster without weakening the quality of its legal decisions. Ideally, they should improve both, which is why at DraftPilot our features around embedding contract playbook governance and granular user permission sets have become such a differentiator.

Here’s a simple way to test the business case for an AI product. Draw a line down the middle of a page.

On the left, write down what changes inside legal by adopting a tool: hours saved, external spend avoided or increased number of contracts handled per lawyer.

On the right, write down what changes for the company: a deal closes sooner, a supplier starts work earlier or a project avoids waiting in the legal queue. Then add how you will check that legal judgment has not weakened.

If the right-hand column is blank, the initiative may still be worthwhile. But it should be judged as a modest efficiency project for the legal team, not something that can materially help the business.

The best tools will do both.

Thanks for being here,

Daniel

CEO at DraftPilot