Hello from the other side
It's no secret that all of us are running out of time
Hey everyone, it’s been a hot minute since I’ve posted anything here. I know, I know—I completely blew past my grand plan to kick off the “Build With Me” series for open-source e-discovery at the start of the year. Sorry about that.
Let’s just say the last two months have been absolutely monumental. I figured I’d ease into a daily groove, but then life and work in and out of Elefant hit me like a freight train.
I learned a ton—some of it thrilling, some of it humbling—and came away realising I should tweak this newsletter. Here’s the new plan:
More frequent posts, but shorter. Because nobody wants to read a magnum opus every time.
Stop banging on and on about how AI is coming for the legal sector. Guess what? It’s here. The future’s unevenly distributed, sure, but I’m bored with shouting “it’s coming” like a broken record. It’s here. Your problem if you’re still being stubborn about it!
Benchmarking is super important, but you need to be actually working with the technology. It’s so crucial that we are building our own in-house suite for benchmarking, but importantly - can you imagine going on and on and on and on and on about benchmarking typewriters, after clear but imperfect demos of them, when you’re still using a quill?
Also, I’m feeling disillusioned with the endless LinkedIn hot takes (no offence to the truly good writers out there), so I’m just going to focus on building stuff, letting the results speak for themselves, and pulling in all who are interested in getting their hands dirty, rather than just hearing from another pundit.
Real Talk About the AI Engineer Conference
I should mention that about two weeks ago, I attended the AI Engineer Conference in New York City, and my mind was blown in the best way. For once, it felt like a place to talk about actual building—not just schmoozing. On day one, I had a coffee chat with Max, the founder of Monadical, when I gravitated towards the pantry, wanting to warm up from the subzero weather with a hot (and to my horror, leaking) cup of coffee. I ended up in this incredible hour-long chat with other founders and engineers that reminded me why I love working in tech. For an alleged “imposter,” being able to follow along the discussion felt great. I didn’t say much at all, which was weird, but I was cold, and also absorbing a ton of information.
Plus, I snagged new tips on code editors, new debugging tools, new everything. It hit me again how passionately engineers share knowledge—whereas in law, we kind of sit around waiting for a big brand to drop a million-dollar solution on us. At the risk of sounding like a broken record, we need to change that.
Tools that I’ve gotten my hands dirty with:
L4, the Domain-Specific Language for law being developed by a team from the Singapore Management University’s Centre for Digital Law. The future really is here, along with other Rules-as-Code projects such as DocRef, and Blawx.
The new tools in Cursor
n8n (shoutout to Ivan Rawter of Pactly for pointing it out)
The various new prompting methods and workflows for Chain-of-Thought/thinking models and research agents (or agents abstracted behind agents rather).
Updates from Elefant
Inside Elefant, we expanded coverage to four more jurisdictions—meaning a whole bunch of new code hurdles, and the scramble to set up thorough quality/assurance processes. We’ve been busy with welcoming our pilot users and beta testers, graph databases, retrieval-augmented generation, data extraction, and data enrichment. And here’s the best part: you don’t need a sky-high valuation to push the envelope. I honestly question the puffed-up startup valuations, even as I build a start-up.
I keep seeing lawyers fixated on the legaltech with the biggest founding rounds, and I’m over here going: why? And no, this is not coming from a place of jealousy. Such startups have their role in the ecosystem, and they can really push the envelope when they bring to bear the gargantuan amounts of resources. It’s a whole other question when it largely goes into sales.
But let’s also not forget the small, but efficient, and sometimes, even beautiful. Deepseek comes immediately to mind. So does my home, Singapore, and other city-states that punch above their weight.
Programmers build their own tools or pick them up from the open-source community all the time. Carpenters geek out about different chisels and saws, but couldn’t care less if the workshop got its money from Whale-Sized Venture Capital. They care about craft. So why are lawyers so enthralled by brand-name vendors, and funding rounds? I think it’s telling when legal practice cleaves very cleanly into Biglaw, and then everyone else. Again, as I’ve said before, my animus lies not against Biglaw practice itself, but all the various unsustainable facets of living that accompany it.
New Angle to eDiscovery
Now, coming back to my #BuildWithMe series, I ran a Deep Research prompt (link to whole chat here) on e-discovery pain points and legal requirements in the US, UK, and Singapore—then realized I bit off more than I could chew. I won’t go through the reasons in detail - you can peruse the prompt if you have the time; Deep Research’s depth and range of sources is something to keep watch of - but it’s time for a narrower focus:
When a client drops a gazillion files on you, how do you get up to speed—fast?
Technically, that’s not “pure” e-discovery, which is more about preservation, proportionality, and figuring out which docs to actually disclose. Plenty of effective vendors already handle that. Instead, I’m talking about a quick ramp-up to understand the core of your client’s case. We have better tools now for that. Let’s stop pretending otherwise and just start building them.
Next week, I’ll share the practical steps and code to do exactly that. We’ll use Claude, some vibe coding, and maybe even brew some ginger beer. (Kidding. Sort of.)
Coming Soon & One Small Favor
I’m teaching a class later today on the Modern Legal Office—our biggest turnout yet. If you’re not already set up with decent data governance processes and ways to handle your documents, the AI wave will leave you in the dust. It’s a pivotal moment.
Also, if you’d rather listen to this as a podcast, let me know. I’m playing with new formats—speaking out loud, transcribing, letting AI tighten up the grammar (the ideas remain mine, though).
I’ll share more about Elefant soon, but for now, keep an eye out for next week’s post.
We’ll take a good, hard look at how to handle that massive doc dump from your client, without losing your mind—or your practice area. Lawyers can build cool stuff, too. We just need to start doing it.
And if you think I’m wrong (or if you agree and want to push this forward), hit me up! I now consult on this kind of thing.
Alright, that’s enough soapboxing for now. The future’s here, and we might as well get our hands dirty.