We have been working with the consumer-grade frontier models to develop what we call "lexploits" in legaltech, and they are insanely good at finding bugs across integrated pipelines. They're also surprisingly good at mitigating them!
Security vulnerabilities are one thing, but in legal we offer up a concept of "knowledge security" which goes to protecting the fidelity of the agent's legal context. Software bugs seem much more tractable because they're managed by software engineers, as opposed to the pipeline "vulnerabilities" we're finding. We wrote a little about one vector here where legal documents aren't quite what they seem: https://tritium.legal/blog/noroboto
No doubt there are many such knowledge domains exposed today. These are more concerning because they're understaffed and managed by non-technical people for the most part. No Mythos required.
We're dealing with malicious fonts in legal contexts, too. There, the human-visible font tells a different story from its Unicode / machine interpretation in documents like PDF and DOCX[1]. Others have considered the same with web fonts and agents. It's concerning to consider how far things might go if you string together a few exploits and couple them with a binding legal obligation. Or worse, an immediate, irreversable payment.
I love this video. It's classic Steve Jobs in a real meta way.
While I agree with the thesis, the response is total reality distortion field.
He says "you have to start with the customer experience" rather than the technology.
Then he name drops 4-5 technologies that were speculative endeavors and says when Apple put them all together to make the laser printer: "we can sell this".
Still in the uncanny valley for me. Like watching AI the film. That said, it’s 3-minutes long and maintains the setting across many different angles, zoom levels, etc. pretty impressive.
Man in building Tritium[1] I have always used the analogy that developers would never program in a web-based IDE. Thus, lawyers would never live in a web-based legal IDE either. In exchange for that we’ve paid the onboarding price of trying to get desktop software installed to even run a demo. This is super timely to push us back towards a reality that web may be viable.
Hi Drew, I remember your "Show HN" from a while back and have been secretly rooting for you ever since! (I'm not a lawyer but for some reason I have many friends that are, and now I happen to do work for a firm in the legal publishing sector, so I often hear about how terrible "word processing" can be and think there've got to be better tools!)
May I ask, how are things going? Also, will your IDE always be focusing on transactional law or have you considered expanding to other legal areas and/or markets?
Hi! It's a super interesting time to be in legal tech. Thank you for asking.
When this project got started, "VS code for transactional lawyers" was the target. We pretty well have that on offer at this point, but it sits in a weird spot making it harder to sell than it would be in, say, 2024. Right now, "AI forward" lawyers are spinning out of law firms in droves to start "AI native" firms backed for example by YC. They're so comfortable with Claude that they for the large part bypass a need for Tritium (or at least they think they do ;). OTOH, large law firms are inundated with legal tech products right now and have a hard time even understanding how an IDE benefits their lawyers. We're also trying to stay away from VC funding (other than from a certain awesome one ;), so we're missing a key signal for enterprise buyers. As I mentioned above, it's super hard to even set up a hands on demo because we have to get the desktop app installed on their infrastructure. But I'm shocked to learn that Googlers are happy to work in a browser, and distributing Tritium via browser is trivial, so we're going to 180 on that right here and now.
That all said, we eliminated the "free tier" as advised back in the Show HN thread, and we've managed to find a very small market in individual users. We're also finding some opportunities with the AI natives using an "unreal engine for legal tech" model that makes Tritium source available and handles the boring editor-related parts of their innovation.
I should probably do a post on this, but there's actually a topic we're working on that perhaps the HN audience will find even more interesting... coming soon!
[edit: I realized that I haven't responded to your question re: other markets, but accidentally did with the hint. We have some ideas.]
Wow, thanks so much for taking the time to answer my questions in detail!
> As I mentioned above, it's super hard to even set up a hands on demo because we have to get the desktop app installed on their infrastructure. But I'm shocked to learn that Googlers are happy to work in a browser, and distributing Tritium via browser is trivial, so we're going to 180 on that right here and now.
"Trivial" in the sense you can just compile everything to WASM? I'd be curious to know what such an IDE would feel like in the browser. I think the only WASM-based GUI apps I've tried in the browser were Flutter apps and those were… weird.
> I should probably do a post on this, but there's actually a topic we're working on that perhaps the HN audience will find even more interesting... coming soon!
> "Trivial" in the sense you can just compile everything to WASM? I'd be curious to know what such an IDE would feel like in the browser. I think the only WASM-based GUI apps I've tried in the browser were Flutter apps and those were… weird.
Yes, that's about it. We rely on threads a lot in the desktop version which doesn't map as easily to WASM so there is still some work to do. But if you remember back to the original Show HN post, it was running in the browser there. So we have experience with it.
There is a bit of uncanny valley that comes with using WASM with <canvas> in the browser like we do rather than the DOM. There aren't reflow events in the same way, and frankly it's just a lot snappier than you expect. But it comes with a lot of trade-offs and you're forced to reinvent the wheel if you totally abandon web primitives.
The main thing holding most people back from web-based IDEs is restricted filesystem and tools integrations, but cloud office suites are extremely popular. Google has excellent infrastructure for distributed build and test cycles built into Cider to go along with the entirely remote version control system.
Best of luck on your web-based demos! Dropping people into a working dummy environment with a few tutorial prompts should really help conversions.
I think they just meant that an Electron app is still at least (half) browser-based, just you're shipping the user a particular browser alongside the app code.
I too have been following your talk on Tritium, so that for the past few years as I have been hand-crafting a new open document engine for the web I have had your use cases in mind as ones that I might be able to solve if things went very well. That engine is shipping today and I think things have gone quite well, so I would absolutely love to talk to you about how we could reshape the landscape if we put our heads together. I feel certain that what I've built could be made into a groundbreaking product in the law world, if only I knew anything about law (not from TV)!
Yes, except Electron apps have the ability to break out of the sandbox because they control it. I take the point that the web stack is used in Electron apps, but it's the other issues, esp. as compounded by cross-browser compatibility that are hard to solve in what most people think of when you say "a browser" context.
Welcome to captive audience pricing! There are more a few companies who have this type of business, especially targeting those in institutions of all kinds.
They probably are fully gone now, but when I was in college some (IRL) classes, usually the big auditorium ones, added interactivity in the form of realtime polls and quizzes with a little “clicker” device. This was of course $30 or whatever and just used some custom RF protocol to register your vote across the room. Single-source, you have to buy it to be in the class.
Textbooks themselves, electronic or not, same racket. Professor is sold the book, but it’s the students who pay. (Don’t forget of course the scam of “writing your own math book” and requiring it!!)
Prisons: some private company always has a deal to “supply telephone service” and charges the inmates or their families rates that are higher than international long distance used to cost.
All of these things are sold to administrators who have no fiscal concerns with the service or product because the institution isn’t the one paying, so there’s zero pricing pressure. If there’s even multiple contractors in the niche, they are more incentivized to compete on sending cool freebies to the administrators, or add perks that benefit them, than they are to compete on pricing for the students/inmates/etc. like, say, Jostens might throw in “free school ID cards” which is technically “saving the school money” in order to get the yearbook contract, while making $100 a yearbook in gross profit on $150 yearbooks. Note: all numbers made up.
My issue with this type of thinking is it assumes "transport cost <<< manufacturing cost" -- a decent assumption for a lot of goods throughout a lot of history, but just... not really true for lots of things in a modern supply chain.
The cost of moving the gown between users -- in the form of the user needing to give back the gown to the service, who must then clean it, inspect it, etc. -- may in fact be far higher than the cost of manufacturing a new gown and only needing your supply lines to be "one way".
Sure, but there's a lot of random matter on Earth -- excess trash being an issue is less about space and more about externalities (e.g. toxic chemicals leaching).
Being mindful of how much trash we produce does not necessitate producing less (or more!) -- but merely balancing the pros and cons.
Yes. Competitive forces would push the cost toward the most expensive input which is likely people. That would be somewhat muted if the supplier was sole source but even then outright purchases would put downward pressure on the rental price.
What competitive forces? It's not like people have a choice in choosing whether they want a particular cap or gown and the people who contract the rental agreement (i.e. the university admin) are not the ones bearing the cost.
> But the real danger with these IMO is that they're turning casual conversations into a permanent record, and one that will be completely discoverable in court, should the company get into trouble later.
The parts that aren’t privileged. On the other hand, perhaps the truth-seeking function of the justice system will be better equipped than before when we had to rely on (more) faulty human recollection.
Security vulnerabilities are one thing, but in legal we offer up a concept of "knowledge security" which goes to protecting the fidelity of the agent's legal context. Software bugs seem much more tractable because they're managed by software engineers, as opposed to the pipeline "vulnerabilities" we're finding. We wrote a little about one vector here where legal documents aren't quite what they seem: https://tritium.legal/blog/noroboto
No doubt there are many such knowledge domains exposed today. These are more concerning because they're understaffed and managed by non-technical people for the most part. No Mythos required.
reply