Why I left a law firm to build software for lawyers

People assume there must have been a moment. A bad case, a worse partner, a Sunday night that broke something. There wasn’t. I liked being a real-estate solicitor more than I expected to, and I was reasonably good at it. The reason I left is less dramatic and, I think, more useful: I got tired of being the software.

Let me explain what I mean, because it’s the whole reason I do what I do now.

The day I noticed

Real-estate work has a particular texture. A lot of it is taking information from one place, checking it against rules, and putting a structured version of it somewhere else. Title, searches, leases, certificates. It is high-stakes, because a missed restrictive covenant is a real problem for a real family. And it is, in large part, mechanical - not in the sense that it requires no skill, but in the sense that the skill is being applied to the same shapes over and over.

One afternoon I was doing a report on title for the umpteenth time, and I caught myself working like a particularly expensive lookup function. Read the entry, recognise the category, apply the known treatment, write the standard paragraph, move on. The genuinely hard judgement - the bit a client was actually paying for - was maybe fifteen per cent of the task. The other eighty-five per cent was me being a slow, costly, occasionally fallible piece of automation.

And here’s the part that stuck: the tools that were supposed to help with the eighty-five per cent were sitting unused on the network drive. The firm had bought them. Nobody reached for them. They didn’t fit how the work actually flowed, so they sat there as a line item, and we kept doing the mechanical part by hand because by hand was, perversely, the path of least resistance.

That was the gap. Not “lawyers won’t use technology.” Lawyers will use anything that makes their Tuesday better. The gap was that almost nobody building the technology understood the Tuesday.

Practising first turned out to be the unfair advantage

I want to be honest about why I think I can do this, because it isn’t that I’m a brilliant engineer. I’m not. The advantage is that I spent years inside the work, which means I know things about it that you cannot learn from a discovery call.

I know which steps feel mechanical but actually hide a judgement that will bite you if you automate it. I know which “obvious” automations a fee-earner will quietly refuse to use, because the failure mode is career-ending and the time saved isn’t worth the risk. I know that the real bottleneck in a deal is rarely the thing the software vendor is demoing, and usually the boring handoff between two people that nobody owns.

Most legal software fails not because the engineering is bad but because it’s solving the problem as imagined from outside. It automates the part that was easy to automate, not the part that was painful to do. Having sat in the chair, I find the painful parts faster - and I’m much harder to fool with a demo that looks impressive but solves nothing real.

What I was actually leaving for

I didn’t leave because I’d fallen out of love with law. I left because I’d fallen in love with a specific, slightly obsessive idea: that you could build tools fee-earners reach for because they’re genuinely better, not because a partner mandated them. That the test of legal software shouldn’t be whether it wows a buyer in a boardroom, but whether a tired associate at 6pm chooses it over doing the thing by hand.

That is a much higher bar than most of the industry sets for itself, and clearing it is most of the work. The model was never going to be the hard part - even in 2025 the models are extraordinary. The hard part is the same thing it was when I was the one doing reports on title: changing what a real person actually does, in a real workflow, under real pressure, when they have every reason to stick with the way they know.

What I tell lawyers thinking about the same jump

I get a version of this question most weeks now, usually from a senior associate or a junior partner who feels the same itch I did. So, the honest advice:

Your legal experience is an asset, but only if you treat it as raw material rather than authority. Knowing how the work is done is gold. Assuming you already know how to fix it is the trap - the fix has to be built and tested and very often thrown away, and your seniority earns you exactly zero shortcuts in that process. The market does not care that you were a partner.

Be prepared to be a beginner again. I went from being one of the more competent people in the room to being the least technical person on my own team, and that is a genuinely uncomfortable trade. It’s also the price of building something real, and I’d pay it again without thinking.

And stay close to the work. The pull, once you’re out, is to drift up into strategy and decks and abstraction, where everything is cleaner. Resist it. The whole reason a former lawyer can build good legal software is that they remember the Tuesday afternoon. The moment you forget it, you become exactly the kind of outsider you left in order to out-build.

I didn’t leave law. I just stopped being the lookup function and started building it - so the next person can spend their fifteen per cent on the judgement, and hand the rest to something that was always meant to do it.