Building Legal Agents When You Can't See the Data - Jacob Lauritzen, CTO of Legora

“People were asking us about Mythos and everything, even though I think realistically a lot of legal work does not require Mythos-level intelligence. Paralegals, gone.”

Claim: A lot of legal work does not require Mythos-level intelligence.

“People were asking us about Mythos and everything, even though I think realistically a lot of legal work does not require Mythos-level intelligence. Paralegals, gone.”

Claim: Paralegals are gone.

“Junior lawyers, gone. I don't need them.”

Claim: The speaker says he does not need junior lawyers and that junior lawyers are gone.

“Um, that's actually not what we're seeing. And we had one problem where, like, it wouldn't cite properly, and so we were like, "Next model release solved." It went from like seventy percent to like ninety-nine. It's just like we did nothing, and, like, great.”

Claim: A legal citation problem improved from about 70% to about 99% after the next model release, without the team making changes.

“But if you're doing a, you know, four hundred million dollar M&A transaction, you're gonna want Lagora.”

Claim: The speaker says a party handling a $400 million M&A transaction would want Legora.

“We're just gonna be twenty engineers for the rest of the company's history, and, like, we're gonna beat everyone who's much bigger. And we're a hundred engineers now, and it feels like we're way too small. But it's something that I'm toying with is, like, how much of our product can just be the agent versus all the other stuff?”

Claim: The speaker once imagined the company would stay at 20 engineers throughout its history and beat much larger competitors; it now has 100 engineers and feels too small.

“We're just gonna be twenty engineers for the rest of the company's history, and, like, we're gonna beat everyone who's much bigger. And we're a hundred engineers now, and it feels like we're way too small. But it's something that I'm toying with is, like, how much of our product can just be the agent versus all the other stuff?”

Claim: The speaker is considering how much of the product could be the agent itself rather than other product components.

“I think we are just at the core probably very, uh, humble about what we're doing because we've always been the underdog from the very beginning.”

Claim: The speaker thinks the company is probably humble at its core because it has been an underdog from the beginning.

“No one believed the smallest startup in a basement in Stockholm was gonna, you know, out-compete this huge behemoth that was, like, way ahead of us in the US. So I think we've always been underdogs, and so we've always been very humble.”

Claim: The speaker says the company has always been an underdog and, as a result, has always been humble.

“No one believed the smallest startup in a basement in Stockholm was gonna, you know, out-compete this huge behemoth that was, like, way ahead of us in the US. So I think we've always been underdogs, and so we've always been very humble.”

Claim: People did not believe the smallest startup in a Stockholm basement could outcompete a much larger US company that was far ahead.

“Um, we've always sort of had to fight for our, uh, existence. Um, I think something we've had to learn as we've also become more and more, um... as we've entered the US and been in the US a lot more, is to be a little bit more, uh, bold and, like, s-s-tell the world how great we are 'cause it, it doesn't come naturally to, I think, some of us.”

Claim: As the company has expanded its presence in the US, it has had to learn to promote itself more boldly, since self-promotion does not come naturally to some team members.

“Um, we've always sort of had to fight for our, uh, existence. Um, I think something we've had to learn as we've also become more and more, um... as we've entered the US and been in the US a lot more, is to be a little bit more, uh, bold and, like, s-s-tell the world how great we are 'cause it, it doesn't come naturally to, I think, some of us.”

Claim: The speaker says the company has always had to fight for its existence.

“Yeah, absolutely. The way everything started was ChatGPT with a Swedish loss- ... search tool in it. That was literally the entire everything. And so that's where we started. We started with legal research, which is like a huge market basically in the world where, um, lawyers spend a lot of time figuring out what are previous cases that look like this, what's the legal precedent?”

Claim: The company began with ChatGPT integrated with a Swedish legal-search tool, which constituted the entire initial product.

“... search tool in it. That was literally the entire everything. And so that's where we started. We started with legal research, which is like a huge market basically in the world where, um, lawyers spend a lot of time figuring out what are previous cases that look like this, what's the legal precedent?”

Claim: The company initially focused on legal research, a large market where lawyers spend substantial time finding analogous prior cases and legal precedent.

“Like, is this, you know-- what have previous people done in these situations? That's extremely important. Um, and so that has always been part of our product. Basically, we started from, like, an, an agent you could ask questions, and it would query loss.”

Claim: Legal research has always been part of the company’s product.

“Like, is this, you know-- what have previous people done in these situations? That's extremely important. Um, and so that has always been part of our product. Basically, we started from, like, an, an agent you could ask questions, and it would query loss.”

Claim: The product began as a simple question-answering agent that queried legal sources.

“That was the very simple beginning. Um, and then from that, it's expanded. And so it's expanded in, like, what can the agent do? It, it will now, you know, draft a lot of stuff really well.”

Claim: The company’s agent can now draft many kinds of material well.

“It will edit Word files. It will, you know, move your files away, organize everything, um, run really complex analyses. Um, and we've added just a bunch more features to it.”

Claim: The company’s agent can edit Word files, move and organize files, and run complex analyses.

“So one of our features is called Tabular View, which is, um, now a name that's being used in industry, and we're very happy to see Claude for legal using- ... our terminology for what they're building. Which is essentially, like, a way to just bulk extract and bulk review many, many, many, um, agreements, which you do in due diligence and when you raise money or whatever. So that's part of it. It's like we help lawyers just do their work much more efficiently, much, much better.”

Claim: Tabular View enables users to bulk-extract and bulk-review many agreements, including for due diligence and fundraising.

“Which is essentially, like, a way to just bulk extract and bulk review many, many, many, um, agreements, which you do in due diligence and when you raise money or whatever. So that's part of it. It's like we help lawyers just do their work much more efficiently, much, much better.”

Claim: The company’s product helps lawyers perform their work more efficiently and effectively.

“And then outside of just doing the work, we're also now, um, helping orchestrate the work a lot more than we were just even a year ago. So we have different sort of-- We have a, a thing called Lists, which is s-sort of like Linear for us.”

Claim: The company’s Lists feature serves a role somewhat like Linear.

“And then outside of just doing the work, we're also now, um, helping orchestrate the work a lot more than we were just even a year ago. So we have different sort of-- We have a, a thing called Lists, which is s-sort of like Linear for us.”

Claim: The company is now more involved in orchestrating legal work than it was a year earlier.

“And, um, we have Intake, which is a way to collaborate with, um, clients. And we have, um, Portal, which is also sort of a way to collaborate, you know, lawyer to lawyer, client to lawyer, lawyer to agent, agent to agent, maybe.”

Claim: The company’s Intake feature enables collaboration with clients.

“And, um, we have Intake, which is a way to collaborate with, um, clients. And we have, um, Portal, which is also sort of a way to collaborate, you know, lawyer to lawyer, client to lawyer, lawyer to agent, agent to agent, maybe.”

Claim: The company’s Portal feature supports collaboration among lawyers, clients, and agents, including lawyer-to-lawyer, client-to-lawyer, lawyer-to-agent, and possibly agent-to-agent interactions.

“We're, we're spending a lot of time there 'cause, um, there's just frankly, the way that you work with a lawyer ten years ago or maybe even today is, is really painful, and there's loads of room to do that- ... in a much better way, to be honest.”

Claim: The speaker considers the process of working with a lawyer ten years ago and even today painful, with substantial room for improvement.

“And so those are, like, I think the two things. It's, it's doing the work, and it's also, like, orchestrating the work.”

Claim: The speaker describes the company’s product as both performing legal work and orchestrating it.

“So I mean, we've al-- we basically had to switch from the assistant to the agent- ... we call it, right? Um, I guess that the agent, I don't know, I wanna say a year ago probably. Um, somewhere.”

Claim: The company switched from an assistant to an agent product roughly a year ago, according to the speaker’s uncertain estimate.

“Yeah, I was in YC twenty-three. Lagora was twenty-four.”

Claim: The speaker says he was in YC in 2023 and that Legora was in YC in 2024.

“Exactly. That, that's, that's exactly what it is. It's, it's CoWork or ChatGPT- ... just like really, really good for legal work.”

Claim: The speaker considers CoWork or ChatGPT very good for legal work.

“I mean, we're using a, a bunch of different ones- ... depending on task, right? That's, um-”

Claim: Legora uses different models for different tasks.

“I think we found it most effective to subdivide the problem and, and, and route on, like, as we subdivide the task. But we do right now have a, a picker In the agent. Mm-hmm. I'm not sure if that's gonna stay there always, to be honest. Ah. Um, so yeah. Yeah, I mean, so I bring a little bit of context to this because I'm advisor for cognition. Yeah.”

Claim: Legora currently offers a model picker in its agent, though the speaker is unsure whether it will remain permanently.

“Uh, and when I joined, you know, part of my, my thesis for what an agent lab is, an agent lab is a, you know, uh, ideally a domain-specific GPT wrapper.”

Claim: The speaker's ideal conception of an agent lab is a domain-specific GPT wrapper.

“Yeah. Right? Like we- Yeah ... we, we run the, uh, we are the domain experts. We'll pay attention to you. We'll, we'll serve your needs way, way more than the frontier labs, and we'll give you the best of the frontier labs- Yeah, we'll abstract away all the- ... according to our judgment.”

Claim: The speaker's proposed role for a domain-specific agent lab is to provide domain expertise and customer-focused service beyond what frontier labs offer, while selecting the best frontier models according to its judgment.

“Yeah. See, our job is to decide for you because your not, your job is not to- Exactly ... use models. We'll always keep you on the Pareto frontier- Yes ... and make sure that it just, it's great all the time, and you don't have to dive into all the details.”

Claim: The speaker believes an agent provider should choose models for customers, keep them on the Pareto frontier, and hide the underlying model-selection details from them.

“'Cause like, one of the things that I've, exactly to this point, if you go to a lawyer and you ask them, "Oh, you're gonna, you know, uh, draft this contract based on this precedent, do you think that's a Sonnet high or an Opus mid task?" You know, they, they won't know. Just give me the smartest model. I swear, exactly. Just, like, make sure it works. My hourly rate is $1,000 an hour. Exactly. Just like- Make it work.”

Claim: The speaker says lawyers generally cannot identify the appropriate model tier for a contract-drafting task and instead want the smartest model to make the work succeed.

“Um, so yes, I originally came in with that. I was like, "Oh yeah, this is, this is what Devin does differently." And now we have five things in the model picker, including Fable and Opus a- and Astra.”

Claim: The speaker initially viewed the product as differing from Devin, but it now offers five model-picker options, including Fable, Opus, and Astra.

“Yes. And, uh, people wanna experiment with the fancy new model. Yeah. Yeah. I... Do you- That's the- Is that the same? I, I think so. I think it's prob- It's, it's definitely way less in legal than it is in, um- Coding ... engineering.”

Claim: The speaker believes users are less likely to experiment with new models in legal work than in engineering.

“Yeah. But there are definitely people that would just want Fable one right away, and you know, people were asking us about Mythos and everything. Yeah, yeah. Um, even though I think realistically, a lot of legal work does not require Mythos level intelligence.”

Claim: Some users want immediate access to Fable and have asked about Mythos, but the speaker believes most legal work does not require Mythos-level intelligence.

“Yeah. There is- Yeah. So, so exactly, right? Yeah. Like, I think basically y- you're like, "Okay, we'll give you the option. You, you asked for this." Exactly. "But you're gonna run down your credits, like, 10X faster."”

Claim: The speaker says choosing the highest-capability model can use credits 10 times faster.

“Yeah. I would have loved... We, we fought introducing it for a very long time, and then in the end we, we just kind of caved in and we're like, "Okay, if you want this level of control- Yes ... you know, then we'll give it to you."”

Claim: Legora opposed adding a model picker for a long time before eventually offering users that level of model-selection control.

“It's a, I think it's a power distribution. So like, there's probably like 60% of the work- ... we can do really, uh, well right now, and the, you know, from now on it's like a cost and latency thing. But then there's the rest of the, the thing where especially for, like, very complex analysis, we, we still need more intelligence.”

Claim: The speaker estimates that roughly 60% of legal work can already be handled very well, while the remaining work—particularly highly complex analysis—requires more model intelligence.

“... we can do really, uh, well right now, and the, you know, from now on it's like a cost and latency thing. But then there's the rest of the, the thing where especially for, like, very complex analysis, we, we still need more intelligence.”

Claim: The speaker says that for legal tasks models already perform well, further improvement is primarily a cost-and-latency issue.

“And then I think there's also just the work expanse kind of. It's like that's the work that we have right now, but also the work before the lawyer who thus probably expands with intelligence- ... I imagine.”

Claim: The speaker imagines that greater AI intelligence may expand the work performed before a lawyer becomes involved.

“That would've been, yeah, Sonnet 4.6, Opus 4.6 is probably right.”

Claim: The speaker says Legora was probably using Sonnet 4.6 and Opus 4.6 when it began building agents.

“Sub-agent delegation was one of them that we've gotten working really well now for, like, fan out type retrieval queries. Works extremely well for us now. Um, that was difficult before. We basically track... So we have a team that's just doing evals, and so we track all of the models' performance in different practice areas, which is like subgroups of, of legal, and different tasks across all of those.”

Claim: The speaker says sub-agent delegation, previously difficult, now works very well for fan-out retrieval queries.

“Works extremely well for us now. Um, that was difficult before. We basically track... So we have a team that's just doing evals, and so we track all of the models' performance in different practice areas, which is like subgroups of, of legal, and different tasks across all of those.”

Claim: Legora has a dedicated evaluation team that tracks model performance across legal practice areas and tasks.

“And then we can see them move, and it's like usually they improve most stuff, but there's actually a lot of model releases where they improve some and, and regress in others. And that actually happens a lot more than you'd think.”

Claim: The speaker says model releases usually improve performance on most tasks, yet many releases produce gains in some areas and regressions in others.

“Yeah. So like there's, um, there's different just, like, framing stuff. So we have different basically, like, worlds, like RL environments essentially, which is like, here's a... 'Cause, 'cause lawyers work in projects always, more or less.”

Claim: Legora structures its legal evaluations around different reinforcement-learning environments representing project-based legal work.

“So it's like, here's 10,000 file data room. You're doing a due diligence. In the end, a great report looks like this. Like, it contains all these ri- red flags, and, like, it doesn't contain whatever weird, uh, stuff, and, like, we want it to be concise, et cetera, et cetera.”

Claim: In Legora's due-diligence evaluation scenario, a good report on a 10,000-file data room identifies relevant red flags, excludes irrelevant material, and is concise.

“That's kind of how we've built it. So there's different tasks, which is like, part of it is drafting a report. Like, does the draft look good? Does it contain all the right stuff? Is the wording sound, the analysis sound?”

Claim: Legora evaluates legal-report drafts for completeness, wording quality, and analytical soundness.

“There's also the part of finding the red flags, like the analysis before that. So we can kind of like subdivide it a little bit. And, and that's how we run the evals.”

Claim: Legora's evaluations separately assess red-flag analysis and report drafting by dividing legal work into subtasks.

“So yes, it's a mix. We have, uh, engineers, and then we have lawyers, and it's one team. So there's like three engineers and like three lawyers right now, and then we have a pool of what we call legal engineers, which is, um, people that have worked in Big Law for like three or four years.”

Claim: At the time of the conversation, Legora's evaluations team included approximately three engineers and three lawyers working together.

“So yes, it's a mix. We have, uh, engineers, and then we have lawyers, and it's one team. So there's like three engineers and like three lawyers right now, and then we have a pool of what we call legal engineers, which is, um, people that have worked in Big Law for like three or four years. Typically, maybe they have a CS degree, and, like, they're very tech forward, and then they come to us and because they wanna, like, help change, uh, the industry. Are they IT in Big Law, or- No ... are they lawyers?”

Claim: Legora's legal engineers typically have three or four years of Big Law experience and may have computer-science degrees and a strong technology orientation.

“Typically, maybe they have a CS degree, and, like, they're very tech forward, and then they come to us and because they wanna, like, help change, uh, the industry. Are they IT in Big Law, or- No ... are they lawyers? They're lawyers in Big Law. They're lawyers in Big Law. Yeah. With CS degrees. Yes. Oh, my God. It's a Venn diagram that...”

Claim: Legora's legal engineers are former Big Law lawyers, not Big Law IT staff.

“Yeah, exactly. That's like our hardest person to hire for. But they're also such a force multiplier 'cause we can pull on them for the eval stuff and for the product development. It's just like you have a pool of 100 experts that use the product every day and go and talk to clients every day, and also, like, they live this and have lived this, and we just pull on them to do evals, and just, like, help us shape the product, and it is such a, such an advantage.”

Claim: Legora's legal engineers help shape its product and evaluations, an arrangement the speaker regards as a significant advantage.

“Yeah, exactly. That's like our hardest person to hire for. But they're also such a force multiplier 'cause we can pull on them for the eval stuff and for the product development. It's just like you have a pool of 100 experts that use the product every day and go and talk to clients every day, and also, like, they live this and have lived this, and we just pull on them to do evals, and just, like, help us shape the product, and it is such a, such an advantage.”

Claim: Legora's legal-engineer group gives the company access to roughly 100 legal experts who use its product and talk to clients daily.

“Yeah, exactly. That's like our hardest person to hire for. But they're also such a force multiplier 'cause we can pull on them for the eval stuff and for the product development. It's just like you have a pool of 100 experts that use the product every day and go and talk to clients every day, and also, like, they live this and have lived this, and we just pull on them to do evals, and just, like, help us shape the product, and it is such a, such an advantage.”

Claim: Legora finds legal engineers especially difficult to hire, but considers them force multipliers because they support both evaluations and product development.

“Evals, um, grew out of the assistant agent team at some point. I think there was a point up until recently where it was like, honestly, just not really worth it 'cause, like, all models just got better.”

Claim: Until recently, the speaker considered complex evaluations not worthwhile because models were improving broadly.

“Evals, um, grew out of the assistant agent team at some point. I think there was a point up until recently where it was like, honestly, just not really worth it 'cause, like, all models just got better.”

Claim: Legora's evaluations function developed from its assistant-agent team.

“So it was like, do we wanna spend a lot of our time doing really complex evals, or do we just, like, wanna build? 'Cause there was so much to be gained from building a better harness and, like, just sort of leaning into everything's gonna improve, and everything kind of just improved.”

Claim: The speaker previously favored building a better evaluation harness and taking advantage of general model improvements over investing heavily in complex evaluations, because he saw substantial gains in that approach.

“Hmm. But- Yeah, like every two, three months you get just a frontier model. It just like sort... I'm-- Yeah, exactly. Just pass it through, right? Exactly. Yeah. I mean, we had one problem where, like, it wouldn't cite properly, and so it would be difficult to follow, and it was like, next model release solved.”

Claim: The speaker said frontier models were arriving every two or three months and that one model release fixed a citation problem that had made results difficult to follow.

“It went from like 70% to like 99%. It's just like we did nothing, and like, great. Yeah.”

Claim: After a model release, citation performance on the problem rose from roughly 70% to 99% without additional work by the team.

“Yes. Yeah.”

Claim: Legora's use of open models was increasing at the time of the conversation.

“Yeah. The huge part of it is just, um, figuring out what does the legal profession look like in five years, 10 years with all of this stuff.”

Claim: The speaker sees anticipating how the legal profession will change over the next five to ten years as a major part of the work.

“Like, coding is maybe a year or two ahead, so, like we can kinda look to that a little bit, but like what are the primitives? How do people work? How do you collaborate if it's not a 200 thread email with, like, V5 Word files back and forth?”

Claim: The speaker thinks software coding may be one or two years ahead of legal work in adapting to AI, making coding a useful reference for the legal industry.

“Like, coding is maybe a year or two ahead, so, like we can kinda look to that a little bit, but like what are the primitives? How do people work? How do you collaborate if it's not a 200 thread email with, like, V5 Word files back and forth?”

Claim: The speaker identifies redesigning legal collaboration beyond long email threads and back-and-forth Word-file versions as an open product-design question.

“Like, what does that look like? And that's, like, not even necessarily purely AI. That's just, like, build a much better product experience.”

Claim: The speaker says that creating a much better legal product experience is not necessarily purely an AI task.

“And make sure you do it with, like, audit logs and data retention and all of those sort of really important things in regulated industry. And then there's the, the AI part, which is, like, make the work faster and better and, and proactive agents that react to stuff when it comes in so you don't have to, you know, uh, kick off everything.”

Claim: The speaker says legal products in regulated industries should provide audit logs and data retention.

“And make sure you do it with, like, audit logs and data retention and all of those sort of really important things in regulated industry. And then there's the, the AI part, which is, like, make the work faster and better and, and proactive agents that react to stuff when it comes in so you don't have to, you know, uh, kick off everything.”

Claim: The speaker envisions AI making legal work faster and better through proactive agents that respond to incoming events without requiring users to start every task.

“So I think a lot of the work is just figuring out what do we think it's gonna look like in five years. If, you know, model, um, capabilities keep increasing, if price keeps coming down, what could we do?”

Claim: The speaker sees forecasting legal work five years ahead as a key task, conditional on continued increases in model capability and decreases in model prices.

“Can you log in and it's like command center where it's just, like you need to react to these five things that the agents have been doing lately, and, like, review and check and improve the system. Maybe that's where we're going, or is where we're going.”

Claim: The speaker suggests that legal work may move toward a command center where lawyers review a handful of recent agent actions and check and improve the system.

“I should preface this by saying I don't know.”

Claim: The speaker explicitly says he does not know what the legal profession will look like in the future.

“I mean, who knows really? But, um, I think we can look to coding a lot. I think, um, all lawyers are gonna be way more, uh, productive than they were before.”

Claim: The speaker predicts that lawyers will be substantially more productive than they were before.

“They're gonna be able to do a lot more stuff at a higher quality, uh, for, uh, lower cost. I think the profession itself, the work itself, is gonna improve a lot.”

Claim: The speaker predicts that lawyers will do more work at higher quality and lower cost, substantially improving the work of the legal profession.

“It's gonna be a lot more... I think lawyers are gonna spend more of their time doing the stuff that they became lawyers to do. Uh, you know, figuring out what's the right risk profile, which manageable risk, what's like the, all of their judgment, strategic judgment, and, like, working with clients.”

Claim: The speaker predicts that lawyers will devote more time to assessing risk, exercising strategic judgment, and working with clients.

“They're gonna do that a lot more and much less of the, like, trolling through big documents to, like, find the smoking gun or, like, do all this menial, super annoying work that a lot of lawyers have been doing.”

Claim: The speaker predicts that lawyers will spend substantially less time searching large documents for key evidence and performing menial tasks.

“'Cause the, the partnership model, I don't wanna get i- into too much of the details, but, like, the partnership model kind of exists because the only type of capital that these companies have is human capital.”

Claim: The partnership model exists because professional-services firms primarily have human capital rather than other forms of capital.

“So, like, it's not like, you know, us or, like, a software engineering firm or, like, a manufacturer or whatever. You ha- you have physical assets or, like, some sort of intangible assets that, that you accrue over time, and, and they pay dividends to the shareholders. But if you work in an industry like professional services, and it's the same for, like, uh, consultancies and, and whatever else, all of your value is in your people, and you can't really own human capital.”

Claim: Professional-services firms derive their value from people, whom they cannot own, unlike firms whose physical or intangible assets accrue over time and pay dividends to shareholders.

“You know, you, you can't you know, slavery is abolished. You can't own people, but you can rent them. You can pay them money. And so partnerships essentially is a structure where you make the human capital the owners of the company so that they stay.”

Claim: A partnership structure makes its human-capital contributors company owners to encourage them to remain.

“Will that change? Like, I don't... It doesn't have to, but I think what we really wanna do with the lawyers, we actually want lawyers to be able, law firms to be able to build some sort of intangible capital, some durable capital.”

Claim: The speaker wants lawyers and law firms to build durable intangible capital.

“And, like, you can build a system that learns of how you work. A lot of firms right now think about, like, well, there's the X law firm way of doing things, and we can codify that, I think, and we can build a system.”

Claim: Law firms can codify their distinctive ways of working in systems that learn how the firms work.

“And whether that's, like, a self-improving knowledge graph or whatever it is, and, like, precedent files and documents where there's, like, a way of working that produces a really good out- output, and that can be their IP.”

Claim: A law firm's codified working methods—potentially represented by a self-improving knowledge graph, precedent files, or documents—could become its intellectual property.

“So it's, like, a way to codify the human expertise and judgment into a system. That, I think, is... Now I'm just rambling completely, but, like, um, that's something that we want to help empower.”

Claim: The speaker wants to help law firms encode human expertise and judgment in systems.

“And does that change the partnership model then? 'Cause the incentives are different in the partnership model. You just gotta bring more clients, and you don't wanna take away time from bringing in clients to codify your information, basically. 'Cause, like, you get rewarded from, you know, eat what you kill or-”

Claim: Partnership incentives reward partners for bringing in clients, potentially discouraging them from spending time codifying firm information.

“... make this luck step, whatever. Um, that needs to change a little. We need to figure out an incentive structure where partnerships, uh, partners are incentivized to codify things for, to build an asset for the company.”

Claim: Partnerships need to incentivize partners to codify information and build assets for the firm.

“But right now they aren't, 'cause as soon as they leave the partnership, they don't, um, get any dividends from the assets that they built.”

Claim: Partners currently receive no dividends from assets they built once they leave the partnership, weakening their incentive to create those assets.

“So it's, like, a... Yeah. I think the, the, the whole business model is, is, uh... We'll, we'll see.”

Claim: The speaker is uncertain whether the partnership model will change.

“Yeah, yeah. Yeah, it, it's interesting what happens to the partnership model, like, the pyramid. I think generally a lot of lawyers, um, you still need ju- you need the next generation.”

Claim: Law firms still need a next generation of lawyers.

“And at least for engineering, the way I think about it is you have a good engineer today, they're worth a lot more even than they were two years ago 'cause they can produce so much more stuff, like, if they're good.”

Claim: A good engineer is worth much more today than two years ago because the engineer can produce substantially more work.

“And I think it's the same thing in legal where, like, they can actually do a lot of really good work if they're good. So I think it's more that, like, really good lawyers or junior lawyers are worth a lot more.”

Claim: The speaker believes that capable lawyers, including junior lawyers, can produce a great deal of good work and are worth much more.

“But it's even more so.”

Claim: The speaker believes that the increased value of good lawyers is more pronounced now than before.

“Uh, so he wrote the application. He, he w- And my feedback to him was, um, you know, people have tried this a million times. Legal tech has never been a thing 'cause, like there's no incentive for lawyers to, to... Why would they be more efficient? They earn less money. Like, it doesn't make any sense. But I think the reason that lawyers do wanna adopt this is that they understand the gravity of the situation.”

Claim: Lawyers want to adopt legal technology because they understand the seriousness of the pressures facing their industry.

“They know that, like the business model and everything is under pressure, and it's like they need to change 'cause the competitive pressure's so high. The billing model itself I think is also gonna change.”

Claim: Lawyers recognize that their business model is under pressure and needs to change because competitive pressure is high.

“They know that, like the business model and everything is under pressure, and it's like they need to change 'cause the competitive pressure's so high. The billing model itself I think is also gonna change.”

Claim: The speaker predicts that the legal billing model will change.

“We've- we're already seeing that, particularly in the US, that a bunch of stuff is actually being priced on like a project fee and not on an hours fee. Um, particularly the like sort of lower-end stuff on the spectrum, like whatever it is, like small reviews and whatever.”

Claim: In the United States, some legal services—especially lower-end work such as small reviews—are already priced by project fee rather than hourly billing.

“Some of it honestly is being done for free.”

Claim: Some lower-end legal work is being provided for free.

“In the strategic work.”

Claim: Law firms make money from strategic legal work.

“Yeah, yeah. Litigation's also... The big work still is there, and it's very valuable. And so they'll, they'll spend-- lawyers will spend more of their time there. Client relationships also super important.”

Claim: Lawyers will spend more of their time on major, valuable legal work.

“Yeah, yeah. Litigation's also... The big work still is there, and it's very valuable. And so they'll, they'll spend-- lawyers will spend more of their time there. Client relationships also super important.”

Claim: Client relationships are highly important to law firms.

“Yeah, yeah. Litigation's also... The big work still is there, and it's very valuable. And so they'll, they'll spend-- lawyers will spend more of their time there. Client relationships also super important.”

Claim: Litigation remains a major and highly valuable part of legal work.

“Like, a lot of law firms will, or historically have basically done stuff for free just to have the relationship with the client.”

Claim: Historically, many law firms have provided some work for free to preserve client relationships.

“Uh, and so now they can do that, but they don't have to lose a lot of money on doing it.”

Claim: Law firms can now provide relationship-building work without losing a lot of money on it.

“Or Lex space has to grow much bigger, I don't know.”

Claim: Legora may need to grow much bigger; the speaker is uncertain.

“You mean the, the super AI built junior legal? Nothing I've found yet. Maybe we should coin one.”

Claim: The speaker has not yet found a name for the AI-enabled junior legal role.

“Legal engineer? That's our and, and also now an industry thing.”

Claim: “Legal engineer” is a term used by Legora and by the broader industry.

“Exactly. Yeah, yeah, yeah. Yeah. And maybe that's, they're those are the people that are, uh, best positioned to do this kind of stuff.”

Claim: People in the legal-engineering role may be best positioned to do this kind of work.

“We do up to a million cells, I think.”

Claim: Legora can process up to one million cells, according to the speaker's estimate.

“And they'll sit there till like 4:00 AM. Yeah.”

Claim: Paralegals may work on document-review tasks until around 4:00 AM.

“'Cause now you can extract it, and you can go and re-prompt everything.”

Claim: Using extraction and repeated prompting can eliminate the need for paralegals to perform some large-scale document-review work manually.

“So like, yeah, I, I think now I don't need a paralegal for that, but like a, a, a good like super AI-pulled paralegal would do something else, you know?”

Claim: For some document-review tasks, a lawyer may no longer need a paralegal, while an AI-enabled paralegal could take on other work.

“Yeah, yeah. They can do more stuff- ... or like spend more time on the things that they, you know, didn't have time for before.”

Claim: AI-enabled paralegals can take on more work or spend more time on tasks they previously lacked time to do.

“Yes, we do that all the time.”

Claim: Legora regularly uses its own product in M&A work.

“No, that works great. We're doing that a lot. Our legal team is, is... I mean, intake has been built with our legal team.”

Claim: Legora built its intake workflow with input from its legal team.

“'Cause right now we have a Slack workflow where everyone throws their stuff in- ... and they're like, "This is annoying." Right? And inefficient. So it's like great. We should use Legora for that.”

Claim: Legora's internal intake process currently uses a Slack workflow for submissions, which employees find annoying and inefficient.

“Yeah, absolutely. So it's, it's, um, we, uh, do a lot of the due diligence ourselves because we can do it, uh, with Tabular Review and other tools. We do still have counsel coming in-”

Claim: Legora conducts much of its own due diligence with Tabular Review and other tools, but still engages legal counsel.

“... um, to like assess certain risk or like, you know, there's payroll. And so Kira was simple 'cause that was all in Sweden, but you know, we've done one in Canada and one in Australia, and so there's like payroll in Australia.”

Claim: Legora engages counsel to assess certain risks, such as payroll in other countries; the Kira transaction was simpler because it was entirely in Sweden, whereas other transactions involved Canada and Australia.

“Okay, do we want someone to look at this like a bit more strategically? So there we will still use lawyers for the strategic part, where they can actually be sparring partners and help us a lot on like how to figure out how to structure this the right way.”

Claim: Legora continues to use lawyers for strategic deal work, including advising on how to structure transactions.

“But a lot of the more menial stuff we could actually just do ourselves.”

Claim: Legora can perform much of the more routine work in its deal process internally.

“So like a lot of, uh, yeah, a lot of the early things we would find, and we have had cases where we found something, uh, that's been like a red flag, and then we've, we've pulled the brakes.”

Claim: Legora has identified red flags during early due diligence and halted transactions in response.

“Well, we're also g- smaller law firms also.”

Claim: Legora serves smaller law firms as well as larger firms.

“Yeah, yeah. I think the European ones, um, have been a bit slower to move. The American ones are very forward, um, move really quickly.”

Claim: Jacob Lauritzen thinks European law firms have adopted AI more slowly than American law firms, while American law firms move quickly.

“I think, uh, it's also very high on their agenda of the American companies, more so than the, uh, European ones, I think. It's like a strategic super important thing we need to just like, do a lot here.”

Claim: Jacob Lauritzen thinks AI is a higher strategic priority for American companies than for European companies, and American companies believe they need to do substantial work on it.

“The American companies generally, both corporates and law firms, also have a much higher appetite to like, custom-build stuff. That's like a thing that they all want to do.”

Claim: Jacob Lauritzen says American corporations and law firms are much more interested in custom-building solutions than their European counterparts.

“Integrations into their own. Like, some of these customers build their own CRM and their own- ... uh, document management system, and their own all sorts of stuff, and they wanna build that, and then they wanna build triggers that like, you do a review in, in, in Salesforce and it goes to this one person for whatever reason, and they do something, and then it goes into Lagoon, and then it goes back in.”

Claim: Some Legora customers build their own CRM systems and document management systems, as well as other tools.

“And there is-- so that's like the simple end, and then there are some that are extremely ambitious, where it's like a whole new product that they wanna build with us, and I, you know, we can't really do that.”

Claim: Jacob Lauritzen says some customers want to co-build highly ambitious, entirely new products with Legora, but Legora cannot really take on those projects.

“Um, but, but that's definitely a thing where like, American companies really want, um, to build their own more than European ones do.”

Claim: Jacob Lauritzen says American companies are more inclined than European companies to build their own solutions.

“Yeah, for sure. I mean, we have- ... great collaboration with all of them. So they'll, they'll say, "This sucks, this is great, we want more of this"- ... "and we'll just build a, a product that does that." Um-”

Claim: Jacob Lauritzen says European customers provide feedback and product suggestions, which Legora uses to build products that address their requests.

“I don't know if that's the reason that they want a custom build versus not a custom build.”

Claim: Jacob Lauritzen is unsure whether American companies' greater interest in custom-built solutions is because they perceive an existential threat to their businesses.

“I think they both know that this is like, uh, extremely important and key f-to their firm's success- ... and their corporate success. So they-- it's, it's both extremely important. I think it's just a different way to achieve it.”

Claim: Jacob Lauritzen thinks American and European companies both regard AI as crucial to their firms' and corporate success, but pursue it in different ways.

“So we have the regular product metrics that we care a lot about. Are people spending a lot of time in our product? Uh, do they seem to get a lot of value out of it? Do they renew it? Do they come back a lot?”

Claim: Legora monitors product engagement and retention through time spent in the product, perceived user value, renewals, and repeat visits.

“We have really good numbers for an enterprise, uh, software of like our DAU over MAU and our like, uh, you know, hours spent per month in the product are extremely high.”

Claim: Jacob Lauritzen says Legora's DAU-to-MAU ratio and monthly hours spent in the product are extremely high compared with enterprise software benchmarks.

“And then we do some sort of-- we do online evals also, where it's like try to figure out generally, like, did we solve the problem here? Just where-- with like a super cheap LLM message, just try to just like have an online metric, and then we can use that also when we do roll-ups of new things.”

Claim: Legora uses a low-cost LLM-based online evaluation to estimate whether a user problem was solved and to assess new feature rollouts.

“We can just like compare feature flag-wise, okay. Does it, did it im-pr-prove or like did it catastrophically decline? Because one thing actually that's might be interesting, that's difficult for our business is we can't look at any traces, basically. We can't look at any data, 'cause it's all super confidential. And that's like extremely frustrating to work with, 'cause you, you basically don't know what people are doing. So it's like here's an agent and it can do so many amazing things, and I have no idea what people are actually doing with it.”

Claim: Because customer data is highly confidential, Legora cannot inspect traces or data and therefore has little direct visibility into how customers use its agent.

“We can just like compare feature flag-wise, okay. Does it, did it im-pr-prove or like did it catastrophically decline? Because one thing actually that's might be interesting, that's difficult for our business is we can't look at any traces, basically.”

Claim: Legora uses feature-flag comparisons to determine whether changes improve outcomes or cause a catastrophic decline.

“I wish. Yeah, I wish. No, but the thing is like a lot of the data that they have is not even, even theirs, right? It's their clients.”

Claim: Jacob Lauritzen says much of the data held by law firms belongs to their clients rather than to the law firms.

“And it's super confidential, so we can't look at it. And so basically, we have an evals team that's like trying to approximate- ... what we think everyone is doing, and like build that into eval sets.”

Claim: Because Legora cannot access confidential customer data, its evaluation team approximates customer use cases and encodes them in evaluation sets.

“We can't re-- There's, I mean, there's always a, a, a, a range, right? Of like if you, if you paraphrase all the way down to like three values and you pick one of them, then obviously that's fine. If you, if you just remove the names, then that's maybe, maybe not fine.”

Claim: Jacob Lauritzen says reducing customer data to a small set of values and selecting one may be acceptable, whereas removing names alone may not be sufficient to make its use acceptable.

“And so like we are, um, generally pick the, the most safe way. So like we, we restrict ourselves as most possible 'cause we, you know, there, there should- ... be no risk at, at all.”

Claim: Jacob Lauritzen says Legora generally takes the safest approach and limits its use of customer data as much as possible to avoid any risk.

“Um, but there are things where it's like, okay, we know that the failure modes are probably these ten or 20, so we can, uh, classify it into one of those, and that's- ... a helpful signal. Uh, we can classify it into the ta-types of tasks and the practice areas also, and then like we have an idea. It's something that we're always trying to, to figure out how to do in a smarter way.”

Claim: Jacob Lauritzen says that when Legora knows the likely failure modes, it can classify cases into those categories to obtain a useful signal.

“... a helpful signal. Uh, we can classify it into the ta-types of tasks and the practice areas also, and then like we have an idea. It's something that we're always trying to, to figure out how to do in a smarter way.”

Claim: Legora classifies cases by task type and legal practice area and continues seeking more effective ways to categorize them.

“This-- All of our infrastructure is on Azure.”

Claim: Jacob Lauritzen says Legora runs all of its infrastructure on Microsoft Azure.

“Um, primarily for this reason, 'cause like when we started, law firms were like, "AWS? No." Like, "No way. We don't trust AWS." It's different now. No, but it-- they were like, "Amazon, Google- ... we can't trust them with any of our data." So we basically, we had to host everything on Microsoft 'cause they already have 365. ... review, et cetera, of all of their documents and all of their Word and whatever. So like Microsoft is an approved vendor in their mind, and so we had to host everything on, on Azure, and that's changed over the past few years.”

Claim: When Legora started, law firms generally would not trust Amazon or Google with their data, so Legora hosted its services on Microsoft Azure because firms already treated Microsoft as an approved vendor after security reviews.

“So like now people understand, okay, we need to get on cloud. We need to do all this kind of stuff. Um, there was also a lot more people talking about on-prem back then. That's also disappeared completely. People understand that like you should just do it in the cloud and it's, and it's safe and it's probably safer than what you're doing.”

Claim: In the period discussed, law firms increasingly accepted cloud adoption as necessary, while discussion of on-premises hosting had largely disappeared.

“So like now people understand, okay, we need to get on cloud. We need to do all this kind of stuff. Um, there was also a lot more people talking about on-prem back then. That's also disappeared completely. People understand that like you should just do it in the cloud and it's, and it's safe and it's probably safer than what you're doing.”

Claim: The speaker says cloud hosting is safe and probably safer than a firm's existing approach.

“Um, so that's changed a lot, but it's extremely painful.”

Claim: The transition in law firms' approach to cloud hosting has been extremely painful for the speaker's organization.

“Uh, yeah. Yeah. I think, I think we needed the AI, uh- ... kick in the butt for everyone to understand that we need to change- ... our ways a little bit.”

Claim: The speaker believes the arrival of AI was needed to prompt the legal industry to change its practices.

“Yeah. The-- We're just thinking about, uh, to, to your point, um, just bring our clients along on like the Pareto frontier. And like the, on the, the map of all the models and what they're good at and their latency and their cost and their, their governance and, and, and their, um, regional availability.”

Claim: Legora aims to help clients choose among models along a Pareto frontier by optimizing for model capability, latency, cost, governance, and regional availability.

“How do we just optimize across all of that? So- Some open weights models are great for certain tasks, and then we'll use them there. Um, some we'll post train and, and some we'll just use the frontier models.”

Claim: Legora selects models by task: it uses open-weight models where they perform well, post-trains some models, and uses frontier models for other tasks.

“Yeah, I think not overall good. Gemini, one of the early Geminis- ... is, uh, yeah, extremely good for certain types of, um, uh, transcription- ... and speaker attribution. So we use that for transla- um, transcription actually.”

Claim: One of the early Gemini models performs extremely well on certain transcription and speaker-attribution tasks, and Legora uses it for transcription.

“Like a Granola? We don't have-- We actually probably will build, like, a meeting thing. But this is just like talk to speech, speech to text- ... text to speech, just like in, in, um, any wor- any other app that has this. And we do actually also transcribe files. If you upload hearings or whatever- ... we'll, we'll then, like, transcribe them, put them into text, and so we can do all of our analysis on it.”

Claim: Legora does not currently have a meeting product but probably will build one; its existing speech feature supports speech-to-text and text-to-speech, and uploaded files such as hearings can be transcribed into text for analysis.

“No Eleven Lab for DPA reasons. Same thing as, like, yeah, data processing agreement, I should say. So, like, our data processing agreement has very few vendors on it- ... 'cause it's really difficult for us. Exactly. We need to keep it- ... super tight- ... on where any data goes, and so, um-”

Claim: Legora avoids ElevenLabs because of data-processing-agreement constraints and keeps the number of vendors and permitted data destinations in its data processing agreement very limited.

“Yeah. No, absolutely. I should actually, uh, shout out also Microsoft's transcription model. They made one recently. It's better than the old Gemini, uh, model yeah, so.”

Claim: The speaker says Microsoft's recently released transcription model is great and better than the older Gemini model.

“Yeah, exactly. And it's like you have all the governance, and it's super scalable, and it's like, then that's probably good enough.”

Claim: A model with strong governance and scalability may be good enough even if it is not state-of-the-art.

“Uh, but yeah, if you, if, if you've just done this on any large scale, you will pick up tons and tons of transcription errors on any- ... domain that is not like, uh, average day-to-day conversation.”

Claim: Large-scale transcription in domains that differ from everyday conversation will produce many errors.

“Yeah. E- like, I think, uh, Whisper and Eleven Labs still transcribe, uh, Claude Code as Cloud Clo-Code. ... very, very common.”

Claim: Whisper and ElevenLabs commonly mis-transcribe the name “Claude Code” as “Cloud Clo-Code.”

“Uh, so, so like that's, that's how you know there's like a lot more to improve.”

Claim: Common transcription errors indicate that transcription systems still have substantial room for improvement.

“... yeah, and our Word add-in, um- ... is also something 'cause that's-- p- a lot of people work, work inside of Word, and so our Word add-in is, is a huge, and, and Outlook add-in also.”

Claim: Legora offers Word and Outlook add-ins; the Word add-in is important because many people work in Word.

“Um, so they-- It's typically, I mean, a good example is just a due diligence, right? Where it's like we need to review all employment contracts, all the supply contracts, all of whatever.”

Claim: A due-diligence review may require lawyers to examine all employment contracts, supply contracts, and other relevant contracts.

“Each row is- ... one employment contract. And then for cells, you ask questions basically. So like the first question, the first, uh, no, sorry, per column. So the first column might be what's the termination, uh, period?”

Claim: Legora's Tabular Review represents each employment contract as a row and uses columns for questions about the contracts.

“... one employment contract. And then for cells, you ask questions basically. So like the first question, the first, uh, no, sorry, per column. So the first column might be what's the termination, uh, period? And then it's gonna like fill out for everyone. It's 30 days or whatever it is. And the next question is like, uh, is there an IP assignment clause or whatever? Next one is like, uh, how much, you know, how much do we have to pay them if we fire them or whatever it is.”

Claim: In Legora's Tabular Review, a question column such as one asking for a contract's termination period is populated across all contracts.

“And then it's gonna like fill out for everyone. It's 30 days or whatever it is. And the next question is like, uh, is there an IP assignment clause or whatever? Next one is like, uh, how much, you know, how much do we have to pay them if we fire them or whatever it is.”

Claim: Legora's Tabular Review can ask across contracts whether they contain an IP assignment clause and what payment is due upon firing an employee.

“Like how much are they getting paid? And then it's gonna fill out the entire thing. You can also just ask the agent to do this. Like if you just drop 10,000 files into the agent and say like, "Do a due diligence on this," it will spin up a Tabular Review and run all of them and wait for it to be done, and then analyze the results, and it can then highlight like red flags.”

Claim: Legora's agent can process 10,000 files for due diligence by creating and running a Tabular Review, analyzing its results, and identifying red flags.

“Okay, this, the CEO is getting paid an outrageous amount, or there's like this weird bonus structure. Okay, then, like obviously we need to look at that. And then the agent from that can draft a memo to send to the client.”

Claim: Legora's agent can draft a client memo based on due-diligence red flags, such as unusually high CEO compensation or an unusual bonus structure.

“Or, uh, it can just do the review and then have, you know, a lawyer or the user just be like, "Okay, let's, let's look at this. Here are the red flags." It's also collaborative so that you can comment on it, um, for other people to look at, et cetera.”

Claim: Legora's review can be presented for a lawyer or user to inspect, with commenting and collaboration features for other reviewers.

“Yeah. It's just a lot better to-- It's, it's super easy to work with, work in. Uh, it's really easy to get an overview super quickly of what's going on, uh, and to review it.”

Claim: The speaker considers a spreadsheet format easy to use and effective for quickly understanding and reviewing results.

“So all of the stuff that we produce also has citations back to source documents, and so you can basically review it super easily. If you wanna just make sure that everything's right, you can also like highlight cells.”

Claim: Legora's outputs cite their source documents, and users can highlight spreadsheet cells to support review.

“These are reviewed or whatever. These are the ones that are flagged, and you can comment and collaborate. That's much easier than any other form factor that we've-”

Claim: Legora's spreadsheet interface lets users mark cells as reviewed or flagged, comment, and collaborate; the speaker considers this easier than other formats.

“People are happy to live there. Uh-”

Claim: According to the speaker, users are willing to work within Legora's spreadsheet instead of needing to export their work to Excel.

“We do use ag-Grid.”

Claim: Legora uses ag-Grid for its spreadsheet interface.

“Yeah, they will. So our type of review actually will run the first 10 rows- ... twice. Yeah. Does this make sense? Did it, did it show us what we thought it was gonna show us?”

Claim: Legora's review runs the first 10 rows twice before proceeding with the full review.

“That's a, that really good-- Oh, but also for people, when people built this, it's just really good for them to iterate quickly on the first 10 rows until like they have a, a-- They're getting the results that they want, and then they can go ahead.”

Claim: Testing the first 10 rows enables users to iterate quickly until the results meet their expectations before running the broader review.

“It's-- No, it's not. I think we have some primitives that work really well. So like Lists is a great primitive 'cause it's essentially, it's like extremely flexible. It's like a Notion- It's like a Notion database, and so you can basically configure it for whatever you want. And we, of course, invest a lot into making this super scalable and work really well. And you can have triggers on them, automations on them, and it's like a great glue to glue everything together.”

Claim: Legora’s Lists feature is a highly flexible, configurable product primitive.

“It's like a Notion database, and so you can basically configure it for whatever you want. And we, of course, invest a lot into making this super scalable and work really well. And you can have triggers on them, automations on them, and it's like a great glue to glue everything together.”

Claim: Legora’s Lists feature supports triggers and automations that can connect different parts of the product.

“It's like a Notion database, and so you can basically configure it for whatever you want. And we, of course, invest a lot into making this super scalable and work really well. And you can have triggers on them, automations on them, and it's like a great glue to glue everything together.”

Claim: Legora invests heavily in making its Lists feature scalable and reliable.

“'Cause then you can kick off an agent run based on a list if something happens- ... and then that comes back and populates a field or whatever, adds new documents or whatever it is. And so it's like, just like you might coordinate a bunch of coding agents with Linear or some sort of task management tool, we can do it similarly with Lists.”

Claim: Legora Lists can trigger agent runs whose results populate list fields or add documents.

“... and then that comes back and populates a field or whatever, adds new documents or whatever it is. And so it's like, just like you might coordinate a bunch of coding agents with Linear or some sort of task management tool, we can do it similarly with Lists.”

Claim: Legora’s Lists can coordinate agents in a way analogous to task-management tools such as Linear coordinating coding agents.

“And so like agents can create lists, and lists can create agents, and- ... agents can read lists, and so it's like the agent swarm comes alive.”

Claim: In Legora’s system, agents can create and read lists, while lists can create agents.

“We need the, the flexibility, even if we don't actually expose it to the end user. Let's say this was just all internal. Then it's a really fast way for us to utilize people like, people like our legal engineers to basically build a lot of products extremely fast.”

Claim: Legora values internal system flexibility, even without exposing it to end users, because it enables legal engineers to build many products quickly.

“So like you can build custom like fund management dashboards. You can basically build that on List with like your own generative UI, and then you basically have a new product that some people are gonna love, and we can have a marketplace of just like a bunch of sp- specialized products that we can scale super quickly 'cause all of our legal engineers can just build new ones.”

Claim: Legora can use Lists and generative UI to create specialized products such as custom fund-management dashboards, and scale them quickly through legal engineers building new products.

“If they're out with a client, "Oh, you want your own? Easy. Like, we'll just fix one for you." So it's like a lovable inside Lagora a little bit, and if we don't wanna expose it, that's also how we build our own custom features.”

Claim: Legora can tailor products for clients and use the same internal approach to build custom features.

“So we, we made another acquisition in the UK called Wexler, and they're doing fact intelligence, which is for litigation mostly, which is like you upload a million documents, uh, for a lawsuit or whatever, and then we extract facts from those with like named entities and dates and whatever.”

Claim: Wexler’s litigation fact-intelligence system can extract facts, named entities, and dates from large document sets, including a million documents for a lawsuit.

“So we, we made another acquisition in the UK called Wexler, and they're doing fact intelligence, which is for litigation mostly, which is like you upload a million documents, uh, for a lawsuit or whatever, and then we extract facts from those with like named entities and dates and whatever.”

Claim: Legora acquired Wexler, a UK-based company focused mainly on litigation fact intelligence.

“And then when you have all the facts, you can basically hold them up against each other and see which ones are corroborated and which ones are, um, what do you call? Disagreement. They-”

Claim: Wexler’s fact-intelligence system can compare extracted facts to identify corroboration and disagreement.

“... you know. And that's extremely powerful 'cause then you can also transcribe a court hearing and you'll be like, "Oh, he said this thing, but actually the email over here says that he was not there."”

Claim: Wexler’s fact-intelligence approach can compare a court-hearing transcript with documents such as emails to surface contradictions about events, such as whether a person was present.

“And that we can also build on this primitive. And so like we can build a lot of really cool things on this one primitive, basically. That's how we think about it, 'cause like we have to be-- As we scale the company, we just have to solve so many things, and like, so, so I think a lot about how do we build a core primitive that we can scale to, to many different end products and create very customizable-”

Claim: Legora can build multiple products on a shared core primitive.

“And that we can also build on this primitive. And so like we can build a lot of really cool things on this one primitive, basically. That's how we think about it, 'cause like we have to be-- As we scale the company, we just have to solve so many things, and like, so, so I think a lot about how do we build a core primitive that we can scale to, to many different end products and create very customizable- ... experiences.”

Claim: As Legora grows, it aims to build core primitives that can support many end products and customizable experiences while addressing the company’s expanding needs.

“Lists is a core one.”

Claim: Lists is one of Legora’s core product primitives.

“Super configurable one. Yeah, projects. It's basically containers of files, file systems.”

Claim: Legora’s Projects feature is a highly configurable system of file containers or file systems.

“Um, the agent, of course, you can kick that off and run that in very different configurations.”

Claim: Legora’s agent can be launched and run with different configurations.

“No, we don't. No.”

Claim: Legora’s agent does not currently have a name.

“Um, people used to talk about Leia as like a, a named entity. Um, but yeah, we s- we stopped that.”

Claim: Legora previously used the name Leia for a named entity and later discontinued it.

“It might be confusing. I don't know.”

Claim: The speaker thinks the name Leia might be confusing.

“Uh, Tabular Review. Which, like we're considering, is that actually an instantiation of-- Like, you can take this as far down as you want, right? But those are Probably the big, the big ones, yeah.”

Claim: Legora identifies Tabular Review as one of its major product primitives while considering whether it is an instance of a more general primitive.

“So we hired our first forward deployed engineer, that was our first hire in New York, also YC, uh, founder, nine months ago maybe. I'm not actually sure. Um, six months ago.”

Claim: Legora hired its first forward deployed engineer in New York approximately six months before the conversation; the speaker first estimated nine months and then corrected that estimate to six months.

“True engineers. But we do-- it, it is something that I think about is, like, how do we exactly enable that at scale? Um, and I think a, a really core thing is, like, how can we empower our legal engineers?”

Claim: Legora is considering how to scale its forward deployed engineering approach, with empowering legal engineers as a core part of the solution.

“Like, how can we power our power users and our legal engineers to customize the product in a powerful way? And then, of course, we, we'll still have forward deployed engineers, like, help out, you know, do all the custom integrations and the, the, the scoping and figuring out what's possible and feed all that back into core product development.”

Claim: Legora aims to let power users and legal engineers customize its product, while forward deployed engineers handle custom integrations and scoping and relay what they learn to core product development.

“Where, it's like, legal engineers should be able to customize a lot of stuff and build a great experience, and then where they can't or we need to make core product changes or custom integrations or whatever, we have forward deployed engineers who also then cycle in a little bit behind, like, sometimes they'll just work on the product, and sometimes they'll work- ... with clients.”

Claim: At Legora, legal engineers should customize much of the product and build client experiences, while forward deployed engineers address needs beyond that scope, including core product changes and custom integrations.

“Yeah. So it's very different client to client, right?”

Claim: Legora's approach to client work by legal engineers and forward deployed engineers varies substantially across clients.

“But, um, if it's a big client, and they're super engaged, and they, like, want-- they have all these dreams of things that they would want to do, or they already have, typically, they have already tried to build something, and it's failed, and they're like, "This is what we-- This is a problem that we'd like to solve," um, then we'll, we'll basically staff them with legal engineers and a forward deployed engineer, potentially, depending on what, you know, their thing is, and then we'll just be like, "Hey, these guys will come.”

Claim: Legora may staff a large, highly engaged client with legal engineers and possibly a forward deployed engineer when the client has ambitious workflow goals or an unsolved problem it has previously tried to address.

“They'll make you successful. They will sit with all your lawyers and teach them how to use it and, like, figure out all the workflows that you have right now and, and build them into Lagora." And some of that will just be a legal engineer, and some of that will require a forward deployed engineer.”

Claim: Legora's legal engineers help client lawyers learn the product and translate their existing workflows into it; some client work also requires a forward deployed engineer.

“Legal engineers configure the product as a power user would. They don't write code.”

Claim: Legora's legal engineers configure the product as power users and do not write code.

“And a forward deployed engineer writes code into production- ... basically. I guess that's the, that's the-”

Claim: Legora's forward deployed engineers write code into production.

“Regular engineers. Yeah. Founders, regular engineers. They're all-”

Claim: Legora's forward deployed engineers have regular engineering backgrounds, and some are founders.

“Yeah, exactly. Dee was our first, yeah She's great. Yeah, yeah, yeah, yeah. She's, yeah, first, first boots on the ground in the US. Yeah.”

Claim: Dee was Legora's first forward deployed engineer and its first person on the ground in the United States.

“Um, some core workflows everyone have, and we can just build in Lagora, and that could be anything from, like, um, "This is how we handle these types of NDAs or whatever."”

Claim: Legora can build common client workflows directly into the product, including workflows for handling particular types of NDAs.

“And then it's just like, uh, you activate it, and it's more, like, skill building, really, for Elise. And most of the FDs, that's more like there's something that we can't do yet, like, we need some more flexibility in some of the primitives or whatever it might be, or we just have this really complex system that we need to integrate into.”

Claim: Legora's forward deployed engineers may be needed when the product lacks necessary flexibility in its primitives or when a client needs integration with a complex system.

“And then they'll go out there, and they'll build the integration against our APIs and maybe expand our APIs or whatever to make that work.”

Claim: Legora's forward deployed engineers build client integrations using its APIs and may extend those APIs when needed.

“So they don't touch the data still. So they'll just build the system. Um-”

Claim: Legora's forward deployed engineers do not access client data and instead build the system.

“... but they will also be onboarded like an employee sometimes. So, like, they might get a laptop and everything, um- ... if they need to. Yeah, yeah. If they need to.”

Claim: When necessary, Legora's forward deployed engineers may be onboarded by a law firm like employees and given a laptop.

“That doesn't happen very often, 'cause, like- ... usually they just build an integration, and it, like, kinda works. But they, yeah, they can if they need to.”

Claim: Legora's forward deployed engineers are rarely onboarded like law-firm employees; they usually build an integration that works without that arrangement.

“It needs to make sense. 'Cause our FDs can also work on core product, right? So it's like it's a constant prioritization thing.”

Claim: Legora allocates forward deployed engineers to client work only when it makes sense, balancing that work against their ability to contribute to the core product.

“No. We, we don't.”

Claim: Legora does not follow a fixed client-site schedule like the four-days-on-site arrangement the interviewer described for Palantir.

“Uh, nothing like that. I, um, to be fully honest with you guys, it's still, it's still, uh, you know, still early days in a lot of this. Yeah.”

Claim: Legora's approach to forward deployed engineering and client work is still at an early stage.

“I do think that it's gonna be more, maybe not exactly like Palantir, but, like, we-- I think we will have a larger in constant circulation group of FDs that maybe sit with a client for, like, two months or whatever to build out this one thing, and then they go to the next one, et cetera, et cetera.”

Claim: The speaker expects Legora to develop a larger, continuously rotating group of forward deployed engineers who may spend roughly two months with one client before moving to the next.

“Yeah, but I, I do imagine us rolling in with legal engineers and forward deployed engineers to, like, work on specific projects.”

Claim: The speaker expects Legora to deploy legal engineers and forward deployed engineers together on specific client projects.

“Um, we do have APIs that you can build against- ... but not the entire surface of the product.”

Claim: Legora offers APIs, but its APIs do not expose the entire product surface.

“So that's been my thing. That's why we've tried not to do it. Uh, we haven't- ... we haven't had APIs. I mean, we started our APIs, like, six months ago or something. We fought it for a very long time, mostly because, yeah, I just don't wanna, uh, we, we move so quickly in, in the product that I don't wanna be, like, slowed down by having to maintain old API versions for, like, a data model that we've changed.”

Claim: Legora began developing APIs about six months before the conversation; it had resisted doing so for a long time because rapid product changes could make maintaining older API versions for changed data models slow the team down.

“I mean, now we're bigger, and things are different, so now it's, it's a cost that we can take on. But, like, when we were 20 engineers, it's just like We, we move so quickly that we can't do this.”

Claim: Legora can now bear the cost of maintaining APIs as a larger company, whereas when it had 20 engineers, its rapid pace made doing so impractical.

“Yeah. I think, I think that's also changing a little bit. A lot of, um, a lot of APIs now, like one thing that we're toying with is like can the agent be the API?”

Claim: Legora is exploring whether its agent could itself serve as an API, reflecting what the speaker sees as a shift in how APIs are conceived.

“I know it sounds weird, but like for some of this stuff, what they actually want to do is they wanna trigger certain things to happen. For some part of this, it could-- you could just be... I'm not saying that we're doing this, but it's like a thing that I'm toying with in my head is could that just be like, okay, you trigger the agent and ask it to do this thing?”

Claim: The speaker is considering whether, for tasks where users only need to trigger an action, Legora could let them invoke the agent to perform it; this is an idea under consideration, not an implemented approach.

“Like obviously it's way less efficient than if you just had like an API that you always hit and it's like, you know. But it's something that I'm toying with is like how much of our product can just be the agent versus all the other stuff?”

Claim: The speaker says invoking an agent for tasks would be much less efficient than calling a dedicated API repeatedly, and is considering how much of Legora's product could consist of the agent rather than other components.

“Yeah, and then I think, uh, deterministic like, um- ... maybe in, you know, a hundred out of, or in two out of a hundred times, it's gonna do something whack that we didn't want it to do.”

Claim: The speaker estimates that an agent might do something unwanted in about two out of every hundred cases.

“I was, I was very strongly the first thing for a very long time.”

Claim: For a long time, the speaker strongly favored limiting scope by saying no to most requests and keeping the engineering team small.

“Let's just say no to everything. Like keep the eng team. I, I think, um, I've said this before, but like one of our presentations that like after I've been at Lago for half a year, uh, I, I pulled up this slide of, um, you know, the three hundred Spartans and, and, and- ... for the movie, and I was like, we're just gonna be twenty engineers for the rest of the company's history, and like we're gonna beat everyone who's much bigger. And we're a hundred engineers now, and it feels like we're way too small.”

Claim: The speaker once expected Legora to stay at 20 engineers indefinitely and outperform much larger competitors; Legora now has 100 engineers, yet the speaker feels it is still far too small.

“I wanted to do that, but I think the, in the real world, we're in like fierce competition. People want so much stuff. It's like, uh, I could be a puritan and just be like, "No, we have to stay twenty people, and we can only build," you know, whatever. Uh, but then we just lose. Uh, and like people would go elsewhere. So we have to build more. And so I think my job is to figure out how do I build a system where we can build a lot, but actually only like where we can... the, the customer feels like we've built so much stuff, but we've actually only built a few things.”

Claim: The speaker believes that fierce competition and customer demand mean Legora would lose customers to competitors if it stayed very small and built little, so Legora must build more.

“Uh, but then we just lose. Uh, and like people would go elsewhere. So we have to build more. And so I think my job is to figure out how do I build a system where we can build a lot, but actually only like where we can... the, the customer feels like we've built so much stuff, but we've actually only built a few things. That's, that's how I think about my job right now.”

Claim: The speaker's current job is to create a system that makes Legora's offering feel extensive to customers while relying on only a few underlying things being built.

“Yeah, exactly. I think that's gonna be... that's really powerful.”

Claim: The speaker considers building with primitives a powerful approach.

“Ooh. We haven't made... I mean, we've made certain strategic bets- ... but we've not tried anything that we don't have a lot of confidence in already. Mostly because we just have like a mountain of demand.”

Claim: Legora has made some strategic bets but has avoided pursuing ideas it lacks substantial confidence in, largely because the company already faces very high demand.

“It's like we want all of these things, and so it doesn't, hasn't really made sense for us to try to be like, "Hey, what, maybe this thing could work," 'cause like there's so many people screaming for the other things. And so we mostly, uh, build those. And then sometimes we'll disagree with clients and we'll be like, "Actually, I think this thing over here will make a lot of sense strategically," and then we'll try to build that anyway. Yeah, exactly.”

Claim: Legora generally builds capabilities requested by clients, but sometimes chooses to build a different capability that it believes is strategically valuable for Legora or the industry.

“Uh, or the industry. So one of those things, um, was, uh, the agent actually. So like no one was actually asking for like, "I want us to be, uh, I want the assistant that I'm using to have more agency and do things without me asking it." Like no one, no one ever. But we're just like, "This def- definitely where the industry is going. This is gonna empower you to do so much cool stuff." Um, collaboration was another one where we're like, we think the industry is going in a place where you can collaborate inside of where you're working, which is Lago, and not in like emails and text messages.”

Claim: Legora believed the AI industry was moving toward assistants that could act with greater agency without users explicitly requesting each action.

“Like no one, no one ever. But we're just like, "This def- definitely where the industry is going. This is gonna empower you to do so much cool stuff." Um, collaboration was another one where we're like, we think the industry is going in a place where you can collaborate inside of where you're working, which is Lago, and not in like emails and text messages.”

Claim: Legora believed the industry was moving toward collaboration within its work platform instead of through email and text messages.

“But they're like some people call them forward deployed legal engineers- ... if you want, but that's essentially what they are. Um, then they go in, yeah, and help, and help them structure their workflows around, uh, the capabilities of the agent.”

Claim: Legora's legal engineers help client organizations structure their workflows around the capabilities of Legora's agent.

“Yeah. So we haven't fine-tuned any models up until now.”

Claim: As of the conversation, Legora had not fine-tuned any models.

“I think now in the past few months, open weight models have become extremely capable, and so we will do it for certain tasks where we think there's a huge delta between what a frontier model can do at the cost and latency versus what we think a, a, a post-trained model can do.”

Claim: The speaker said open-weight models had become extremely capable in the preceding few months and that Legora would use them for tasks where it expected post-trained models to offer a large cost-and-latency advantage over frontier models.

“Legal research is a great example. It's like, uh, you have to make... it's a, it's a fan-out retrieval problem, and you can make a really effective, uh, model to tho- those queries super quickly and super cheaply at low latency.”

Claim: The speaker characterized legal research as a fan-out retrieval problem and said models for its queries can be built quickly, cheaply, and with low latency.

“Very specific bounded tasks Yeah, exactly.”

Claim: Legora post-trains models for narrowly defined tasks, including search.

“Yeah, yeah. We're not trying to like train our own Fable- ... to like be the main agent. I, I don't think that's like a... We won't win that, you know- ... versus Anthropic most likely.”

Claim: Legora is not trying to train its own model to be the main agent because the speaker believes Legora is unlikely to outperform Anthropic in that role.

“He actually talked to Turbo teams to get them to build multi-vector so that we could do these types of things. But no, there's loads of, uh, low-hanging fruit generally even before models. And even in the harness, honestly, like if you just wanna create a better output for our clients, there's like so much stuff that we can do that has immediate impact before training models.”

Claim: The speaker said there are many relatively easy improvements to Legora's harness that can improve client outputs immediately, even before training models.

“Yep. The way that we edit files is important. The way that we do an analysis is important. Uh, the way you do legal research is all baked into it. So if like you might have a problem that comes up or you see something, then it knows how to go and do the legal research to figure out what the answer is- ... what the precedent is. That's all very important. And then of course, there's all the governance stuff around it as well.”

Claim: Legora's legal AI harness incorporates workflows for editing files, conducting analysis and legal research, and handling governance.

“It's, uh ... Yeah, I think it's, it's, it's, um, it's search- ... and analysis. Yeah. It's search, it's analysis, and it's drafting probably.”

Claim: The speaker said Legora's legal AI system involves search, analysis, and probably drafting.

“'Cause like drafting word, drafting huge contracts is actually also not super trivial.”

Claim: Drafting very large contracts is a nontrivial task.

“'Cause you have to ... Let's say you're drafting a huge, uh, contract from scratch, then you have to find precedents of like, how did we do this previously? Do we need this type of clause or not?”

Claim: Drafting a large contract from scratch requires finding relevant precedents and determining which clauses to include.

“They, they figure out what clauses have we used in similar deals and similar can mean a lot of different things, as I'm sure you know in search. It could be like same industry, same type of company, same whatever.”

Claim: When drafting contracts, lawyers look for clauses used in similar deals, and similarity may be based on industry, company type, or other factors.

“Um, and so you need to identify the right precedent. You might have like multiple documents that you then try to figure out whi- which part should I take from what document, and that's actually non-trivial.”

Claim: Contract drafting can require identifying the right precedent among multiple documents and selecting which portions to use, a nontrivial task.

“So it, it's essentially also a search problem.”

Claim: The speaker characterized contract drafting as, in essence, a search problem.

“So I'll say, I was told I was far from this yesterday, uh, definitely Fable and Opus are extremely chatty. I'm sure you guys have known this. Uh, they're very verbose.”

Claim: Fable and Opus are extremely verbose.

“So I'll say, I was told I was far from this yesterday, uh, definitely Fable and Opus are extremely chatty. I'm sure you guys have known this. Uh, they're very verbose.”

Claim: The speaker said Fable and Opus are extremely chatty and verbose.

“I've noticed Astra is fair- pretty good at writing.”

Claim: Astra is fairly good at writing.

“... much more concise. Um-”

Claim: Astra is much more concise than the models being discussed as verbose.

“We, we can ... Yeah, we can- ... we can do, we can do certain things, but I think the chattiness verbosity is actually a little bit difficult for us to do.”

Claim: The speaker's team finds it somewhat difficult to address model chattiness and verbosity.

“So this is, I think, you know, let's see, but probably the next model from Anthropic is gonna be much better at this. Astra is actually very concise.”

Claim: The speaker thinks Anthropic's next model will probably be much better at reducing chattiness and verbosity, and says Astra is very concise.

“Do they make stuff up? So, uh, we have a system of, uh, a citation system where we basically check what they're writing has a link. So the user has a link they can click, and it highlights with a bounding box like in the PDF or whatever where, where this information was taken from.”

Claim: The speaker's team uses a citation system that links written content to a clickable source location highlighted in the underlying document.

“And we can sort of make sure that what it says always is citable somewhere, even multiple layers down. So like if the agent writes something based on a tabular review, you know, c- it can cite the cell in the tabular view or the row in the tabular view.”

Claim: The speaker's system can trace agent-generated statements to citable sources at multiple levels, including individual cells or rows in a tabular view.

“So you can actually like click multiple layers down and find, you know, sort of validate it all the way up.”

Claim: Users can follow citations through multiple layers to validate information back to its source.

“People ship like hallucinated links to Supreme Court, and they don't care.”

Claim: Some people ship hallucinated links to the Supreme Court without caring that the links are fabricated.

“That happens, yeah. Exactly. That happens. Uh, yeah. I, I think it's like coding. You know, you have engineers that just don't give a shit and just like send agents, and you have people that like sit and read every comma.”

Claim: The speaker says review of AI-generated work resembles code review: some people barely inspect agent output, while others scrutinize it closely.

“Yeah, but I think they probably ... It's just like we know in code review what to look for.”

Claim: People know what to look for when reviewing code.

“Like what is the system or whatever. I think they also know, okay, this is the, the thing. These are the red ... Like this is the thing that's important. Let me just like double-check these things.”

Claim: Reviewers can identify the important parts of a system or output and focus their checks on those parts.

“Then like these sound reasonable, so it's probably fine.”

Claim: The speaker suggests that when reviewed items sound reasonable, they are probably fine.

“Hmm. Just like if you, if you receive something from a junior lawyer, it's the same thing you do, right? Just like, eh, I should double-check this, and like this- ... probably is fine.”

Claim: The speaker compares reviewing AI output to reviewing a junior lawyer's work: a person should double-check it, and it may then seem fine.

“Nothing I can think of specifically right now. I think long document reasoning is still difficult. Um-”

Claim: Long-document reasoning remains difficult for models.

“Yeah, exactly. You have a huge contract. Um, and like you're making edits across it. It's a difficult thing to do for a model today even.”

Claim: Models currently find it difficult to make coordinated edits across a huge contract.

“Uh, and like remember that you changed all these things up here, and like the definitions changed and that you, the ordering changed and whatever else.”

Claim: Coordinated contract editing requires a model to track changes across the document, including altered definitions and ordering.

“Um, so the simplest would be one shot. You put, feed everything in and then one-shot output. You don't do that. You break it up into chunks- ... and then you orchestrate edits.”

Claim: For large-document work, the speaker describes a one-shot approach as simplest but says the process instead breaks the document into chunks and orchestrates edits.

“And this is like one of those things that are funny, like that are fun to be part of in this industry, right? So our word add-in was like we, we've done, you know, we've done MapReduce, and we've changed, and there was one big one, and there was like one big one in multiple sites. It's like we keep changing- ... the, the scaffolding that we build around as models change.”

Claim: The speaker's team continually changes the scaffolding around its systems as models change.

“... uh, obviously because of what I do, and I've recently shifted from MapReduce to throw everything in context. And it's, it's better.”

Claim: The interviewer says they recently moved from a MapReduce approach to putting everything into context, which they find better.

“Who'd have ... Yeah, exactly. But it's also, it's different. I mean, we had this problem with search where it's like you're searching for certain things in huge contracts, and you can MapReduce it, but then it might ... If you're summarizing any sort of thing, then it might lose some important nuance- ... in between the chunks, essentially if you just summarize the individual parts up. And so we've ... This has been a problem where it's like we can do two types of search. We can also do the other one where it's like, um, you, you need to have reasoning in, in the different steps there in between.”

Claim: Summarizing a document chunk by chunk may lose important nuance that connects the chunks.

“... in between the chunks, essentially if you just summarize the individual parts up. And so we've ... This has been a problem where it's like we can do two types of search. We can also do the other one where it's like, um, you, you need to have reasoning in, in the different steps there in between.”

Claim: Some document-search tasks require reasoning across steps, rather than relying only on summaries of separate chunks.

“Uh, so I'm trying to think of a really good example here. Yeah, there's, there's some documents where they talk about something at length over like, let's say, two pages or whatever, and then each individual chunk is not necessarily relevant to the search term or what they're trying to figure out. But together it's extremely relevant. And so yeah, these are some of the, you know, problems where it's, uh, been hard.”

Claim: In some documents, information spread across multiple chunks may be highly relevant to a search or question even though no individual chunk is relevant on its own.

“So we have, we have feedback. We can receive feedback from them. And if they- ... are willing to engage- ... yeah, exactly, or they're just like, "Hey, this is like super important to us, and it's not working well." And then we'll be like, "Okay, great. Can you like anonymize the documents?"”

Claim: The speaker's team can receive customer feedback and may request anonymized documents when an engaged customer reports an important problem that is not working well.

“And like- The problem, then whatever, and we'll re- we, if we can recreate it, then we can fix it.”

Claim: The speaker's team can fix a reported problem if it can recreate the problem.

“Exactly. The problem with buying a law firm data is do-- well, you get all the inputs, the, the real inputs, which is great. Uh, the outputs are a little bit more difficult 'cause you just get what their end result was, and it's difficult, uh, to figure out if, like, the way... You have to basically, like, uh, reverse engineer how they got there. So in our product, when we get real traces from real use cases, it's extremely high value. But if we just, like, got this output and we don't know if it's good or bad or how it was created, it's a little bit difficult.”

Claim: Acquiring law-firm data provides valuable inputs, but its outputs are harder to interpret because they show only the final result and require reverse-engineering the process that produced it.

“You have to basically, like, uh, reverse engineer how they got there. So in our product, when we get real traces from real use cases, it's extremely high value. But if we just, like, got this output and we don't know if it's good or bad or how it was created, it's a little bit difficult.”

Claim: Real usage traces from real use cases are extremely valuable for Legora’s product.

“You have to basically, like, uh, reverse engineer how they got there. So in our product, when we get real traces from real use cases, it's extremely high value. But if we just, like, got this output and we don't know if it's good or bad or how it was created, it's a little bit difficult.”

Claim: Legora finds an output difficult to use when it lacks information about the output’s quality and how it was produced.

“Yeah. So basically, don't take default embeddings, even, like, OpenAI embeddings, Gemini embeddings, doesn't matter. Customize your embeddings.”

Claim: Teams should customize their embeddings instead of relying on default embeddings such as OpenAI or Gemini embeddings.

“I think that's right. We-- the only thing is, uh, you need to make sure that you're multi, uh, language. So for us, we, we serve so many languages, so that's been one of the problems.”

Claim: Legora serves many languages, and multilingual support has been one of the company’s challenges.

“I think we need way more model companies out of the-- out of Europe.”

Claim: Europe needs many more model companies, in the speaker’s view.

“Yeah, yeah, yeah, yeah, yeah. I think both. It really depends. I think if you're an, an big enterprise, you might have Claude co-work, and then you'll-- It's-- I think it's similar to, like, how I use Linear through Claude, but I still go to Linear. I think enterprise will do that. I think law firms will just use Legora- ... um, because that's simpler and better.”

Claim: The speaker expects that in two years large enterprises may use Claude alongside existing tools, while law firms will use Legora because it is simpler and better.

“No, I don't think so.”

Claim: The speaker does not expect AI labs to cut off competitors in the legal-services sector as they have in coding.

“Exactly. That's not their call to make. Right? And they don't even know of- like, often what's actually going on. I don't thi- I don't think that's the case. They have so much stuff to capture in knowledge work or in, in, in sort of enterprises knowledge work generally, that, like, legal has to be part of that, just like data analytics has to- ... and sales and marketing and finance. So, like, they have to do things that are good there, but, uh, they're never gonna go as deep as we are. And I think if, if Claude takes a small client, small, uh, you know, corporate clients and they do some of their legal via Claude, like, that's, that's great.”

Claim: The speaker believes enterprise knowledge-work offerings must include legal alongside data analytics, sales, marketing, and finance, but general-purpose providers will never specialize in legal as deeply as Legora.

“... and sales and marketing and finance. So, like, they have to do things that are good there, but, uh, they're never gonna go as deep as we are. And I think if, if Claude takes a small client, small, uh, you know, corporate clients and they do some of their legal via Claude, like, that's, that's great.”

Claim: The speaker considers it acceptable for Claude to handle some legal work for small corporate clients.

“That's fine. But if you're doing a, you know, $400 million M&A transaction, you're gonna want Legora.”

Claim: The speaker expects clients to want Legora for a $400 million M&A transaction.

“Yeah, it's, it's, uh, it's definitely a program that we are-- uh, I think we've started. I actually don't know exactly where we're at, but we, we do want people in law school to- ... learn to use Legora. Um-”

Claim: Legora has begun or is beginning a program to teach law students to use Legora, though the speaker is unsure of the program’s exact status.

“I would actually do it a different way. Uh, you should do scholarships for la- for law school for CS majors, for engineers. We will cover your- Yes We'll co- Yeah, CS plus some software experience- ... we'll put you in law school. We'll cover it.”

Claim: The speaker recommends that Legora fund law-school scholarships for computer-science majors or engineers with some software experience.

“People do JD/MBA. That's a thing. But JD/CS, uh, I haven't seen it.”

Claim: The speaker has not seen combined JD/CS programs, while JD/MBA programs do exist.

“... yeah, engineering, we're hiring in New York, um, Stockholm, and London.”

Claim: Legora is hiring engineering staff in New York, Stockholm, and London.

“Uh, and all other roles we're hiring in everywhere almost.”

Claim: Legora is hiring for non-engineering roles in almost every location.

“100, a hundred.”

Claim: Legora has approximately 100 people in its New York office.

“Um, a lot. I don't... Yeah. It's, it's, it's aggressive.”

Claim: Legora plans to hire many engineers in New York, describing its hiring approach as aggressive.

“Yeah, I mean, like, that, that's the home of corporate law, right? As far as, like- ... uh, mid-America is concerned”

Claim: New York is a major center of corporate law in the United States.

“... so, like, yeah, I mean, I'm looking forward to you buying a law firm and, like-”

Claim: The speaker anticipates that Legora may buy a law firm and says he is looking forward to it.

“Yeah. It, uh- ... it, it's, it's super exciting to, to... I think, like, you know, every industry, especially, like, the top, like, doctors, lawyers, coders, you know, they, they all have to have, like- ... you know, two or three agent labs that, like, just- ... sort of dominate the space.”

Claim: The speaker expects two or three agent labs to dominate each industry, particularly medicine, law, and coding.

“So you guys have, like, a great market. Yeah.”

Claim: The speaker believes Legora operates in a strong market.

“That's what I think. And I think it's only gonna get much bigger as, like, law changes and get-- becomes more, um, available to more people.”

Claim: The speaker expects the legal market to grow substantially as laws change and legal services become accessible to more people.

“Yeah, I know, I know. Which is why I think it's a really exciting thing to work on 'cause, like, modern society runs in some way on legal.”

Claim: Modern society depends in part on legal systems.