Jun 6, 2026 · 57m · news
Inside Legora's Tech Stack: Why Token Maxing is Failing Enterprise Startups | Legora CTO · 20VC with Harry Stebbings
0:00 / 0:00
▶ Watch on YouTube →
gold bands on the timeline = statements, start to end. Hover to read, click to jump. CC turns on captions
In this episode of 20VC, host Harry Stebbings interviews Jacob Lauritzen, the CTO of Legora, to discuss how the hyper-growth enterprise startup leverages AI tooling, restructures engineering workflows, and builds high-performance, ego-free technical teams.
How this conversation actually went
Every chapter scored 0–10 on four independent dynamics. Hover any point for the reasoning behind the score. Harry holds 23.1% of the talking time here. How this is scored →
Harry as informed peer 3.8 Guest teaching 4.1 Guest disagreement 1.7 Harry pushing back 3.1
speaking balance: gold is Harry, purple is the guest (3 minute bins)
Sharpest disagreement ▶ 15:20 Jacob rejects Harry's framing of taste as Silicon Valley rhetoric
When Harry asks if taste is just Silicon Valley BS designed to protect incumbents, Jacob forcefully pushes back, arguing that without explicit taste, AI systems inevitably converge into generic gray slop.
Hardest push from Harry ▶ 37:11 Harry explicitly rejects Jacob's thesis on Cursor's neutralityHarry directly refuses Jacob's argument that Cursor acts as a neutral third-party token optimizer, citing its acquisition by xAI/Grok and arguing model-agnostic competitors like Cognition will win instead.
Biggest teaching moment ▶ 21:18 Jacob delivers a 2x2 framework on build vs. buy software economicsJacob educates Harry on when vibe coding internal software makes sense versus buying off-the-shelf products, categorizing software along axes of surface area and architectural depth.
Harry holds his own ▶ 8:46 Harry cites industry leaders to challenge agentic software determinantsHarry demonstrates domain depth by invoking conversations with Jason Lemkin and Anshini Mitter, challenging Jacob on whether API quality is the ultimate determinant for AI agent selection.
the scores for every segment, with the reasoning behind each
| Chapter | Topic | Harry as informed peer | Guest teaching | Guest disagreement | Harry pushing back | Why |
|---|---|---|---|---|---|---|
| Jacob's Naive Advantage as Legora's CTO | 2 | 3 | 1 | 1 | Harry introduces Jacob and asks whether having no priors in building a large engineering org is a blessing or a curse. Jacob explains that iterating on organizational processes in 2026 requires humility, while detailing their usage of tools like Cloud Code and Cursor. | |
| Shifting Bottlenecks in the Age of AI | 2 | 5 | 1 | 1 | Jacob breaks down the software build cycle into three phases—product scoping, writing code, and reviewing code. He explains how AI compressed code creation, shifting the primary bottlenecks to product definition and code review. | |
| The Nascent State of AI Code Review | 3 | 5 | 1 | 2 | Harry jokes about calling things nascent, while Jacob details how AI code review currently operates with agents arguing with each other. Jacob emphasizes that future human engineers will focus on higher-abstraction systems architecture rather than line-by-line review. | |
| Enforcing Guardrails for Effective AI Agents | 5 | 4 | 2 | 3 | Harry cites conversations with Jason Lemkin and Anshini Mitter regarding API quality as the main determinant for agent selection. Jacob reframes this around guardrail setting and reveals internal model usage stats, leading to Harry probing on security threats. | |
| Evolving Development Processes and Post-Mortems | 5 | 4 | 2 | 4 | Jacob outlines SRE agents running post-mortems and PMs prototyping before involving engineers. Harry challenges Figma's continued relevance, arguing it functions merely as a storage feature. | |
| Taste as the Differentiator Against AI Slop | 4 | 4 | 3 | 6 | Harry directly challenges the term taste, asking if it is just Silicon Valley rhetoric to protect incumbents. Jacob counters that taste represents an opinionated stance necessary to avoid converging into generic AI slop. | |
| Speed of AI vs. Speed of Human Adoption | 3 | 4 | 1 | 2 | Harry jokes about legal adoption speeds, and Jacob highlights the gap between AI development speed and human customer adoption. Jacob discusses internal vibe coding of tools, including an app built to help Canadian hires migrate to Sweden. | |
| Vibe Coding vs. Buying Off-the-Shelf Software | 4 | 6 | 2 | 3 | Harry shares an anecdote of an exec replacing software by vibe coding and asks why companies shouldn't always build instead of buy. Jacob educates Harry with a framework based on surface area and architectural depth. | |
| Career Advice for Next-Generation Engineers | 3 | 4 | 1 | 2 | Harry asks for career advice for new CS graduates, and Jacob emphasizes adaptability and learning speed over static technical skills. They discuss model evaluation across latency and performance trade-offs. | |
| Geopolitics and the Future of Open-Source AI | 4 | 4 | 1 | 2 | Harry brings up geopolitical worries regarding open-source Chinese models. Jacob agrees on the necessity of Western open-source models for sovereign security and discusses context window advancements. | |
| Forward-Deployed Engineers and Customer Education | 4 | 4 | 2 | 5 | Harry asks what competitor Harvey has done better, leading Jacob to admit he drastically underestimated engineering headcount requirements. Harry presses Jacob on whether being understaffed makes them too slow. | |
| Cultural Engineering Differences: Europe vs. the US | 6 | 3 | 3 | 6 | When Jacob suggests Cursor helps enterprises optimize token routing, Harry explicitly disagrees due to Cursor's acquisition by xAI/Grok, arguing model-independent alternatives like Cognition and Factory will win. | |
| The Death of the Traditional IDE | 3 | 5 | 1 | 2 | Jacob predicts traditional IDEs will shift toward visual system architecture design. In response to Harry's self-described naive question about scaling 100x versus 10x, Jacob explains load spikes and fair queuing. | |
| The Talent Dilemma: Elite Engineers vs. Elite Models | 4 | 4 | 1 | 2 | Harry presents a forced choice between six months of superior models or superior engineers, and Jacob firmly chooses engineers. Jacob shares his philosophy on ego-free leadership and stepping aside if someone better emerges. | |
| The Power of In-Person Teams in Stockholm | 4 | 3 | 2 | 4 | Harry asks about maintaining talent bars during rapid hiring in Stockholm. Doing the math, Harry challenges Jacob on whether adding two and a half engineers a week is realistic without compromising quality. | |
| Aquihiring and Integrating Tech Teams | 3 | 4 | 2 | 3 | Harry teases Jacob about buying companies frequently, and Jacob explains that acquihiring is faster than individual recruiting. Jacob discusses giving decisive feedback within two weeks when mishires occur. | |
| The Challenge of Hiring Engineering Managers | 5 | 4 | 2 | 3 | Jacob identifies engineering managers as the hardest role to hire. Harry quotes a cynical management rule to fire anyone on LinkedIn talking about their team, prompting Jacob to explain why technical managers are essential for team health. | |
| Quick-Fire Round and Future of Law | 4 | 4 | 3 | 5 | During the quick-fire round, Jacob claims Legora is the most underrated AI company, which Harry humorously rejects. Jacob predicts lawyers will eventually operate one level above contract text, and they close with a revenue bet. |
Statements from this episode (59)
Legora CTO: AI tooling opportunity cost outweighs token expense
“I don't want to say infinite, but for me, it's a question of opportunity cost. We're in a competitive environment. There's so many things that we can do, and the cost of not doing it is extremely high. It almost outweighs any sort of token cost.”
Lauritzen: Incumbent enterprise giants suffer from a lack of employee enthusiasm
“No one in the 800 pound gorilla is extremely excited to be there.”
Jacob Lauritsen had no prior experience building engineering orgs before Legora
“So essentially I don't have any priors coming into how to build an engineering org.”
Lauritzen: Engineer productivity gains are reshaping software organization structure
“Each engineer can produce much more than they could previously and that just has a ton of ripple on effects throughout the org, basically, and how you structure it”
Lauritzen: Writing code is no longer software development's main bottleneck
“And number two was the primary bottleneck for the past hundred years, almost. So like the rate limiter was how quickly can you write code? That is now super cheap. So that's sort of been compressed. And so the bottleneck now is like the two other ends, which i…”
Legora uses AI code review today despite its nascent state
“We do AI code review today, and it's in its nascent phase.”
Jacob Lauritsen urges founders to build startups solving AI code review
“Current review tools are not good enough. I think we need something new. I keep telling people at all events that I'm at, like, if you're gonna do a startup, please do something that solves the review thing.”
Lauritsen: Code review should focus on architecture and security, not lines of code
“No one wants to look at the, all the lines of code. It's like, what's important is, what are the, what's the impact on systems architecture? What's the impact on systems design, stability security boundaries? How does it sort of take our system in the right di…”
Lauritsen: Software engineers will shift from writing code to system design
“Yeah, I think so. I think so. I think that's right. The job of assist, of a, of an engineer is changing from typing a bunch of code to sort of one layer above it, which is what does the system look like?”
Lauritsen: Legora will soon create a dedicated role for AI agent meta-engineering
“I think the other thing that engineers are doing more and more, which is an explicit role with us soon, which is like kind of the meta engineering of making agents really effective.”
Lauritsen: Enterprise engineers will primarily define guardrails for autonomous agents
“And I think that type of guardrail setting will see everywhere. And so, if you're a big enterprise, and you're rolling out AI tooling, and you have agents that build your own internal software, you know, you have AI tools that build your HRAS system, and your,…”
Lauritsen: Claude and Cursor generate over 50% of Legora's code
“Actually, I took a look recently, and it's Claude and Cursor on the top, and there's like, I think it's like two percent between them. So they are really, really close. And then it's, you know, miles above the next engineer. So they're way above 50%.”
Lauritsen: Legora vendor suffered security incident; predicts more breach events
“One of our vendors had a security incident just yesterday. We just rotated a lot. I mean, just, you know, to be clear, this was internally doesn't affect any of our clients or anything like that, but it's just, I think we're going to see more of them.”
Lauritzen: Legora Uses SRE AI Agents to Draft Incident Post-Mortems
“If you have an incident, now you just unleash an SRE agent, a sort of an incident agent, and it will just super quickly figure out what's going on, look at all the logs, look at all the metrics, metrics, sorry, Telemetry, and it's surprisingly good. It's reall…”
Lauritzen: AI Allows PMs to Test Prototypes Before Involving Engineers
“PEMs can prototype super, super fast. Which is really, really great because that means you can front load a lot of the work. So like a PM can start hyper long before once he or she has the, you know, smallest inkling of an idea that we might want to do this. T…”
Lauritsen: Software teams will skip the design phase for individual feature functionality
“Probably some companies will skip the design phase. I think we can skip the design phase on functionality.”
Lauritsen: Legora still uses Figma for design system consistency and UI standards
“We still use Figma, yes. I know where you're going with this. As soon as you start building a system that's larger than something very small, you want consistency, and you want to have a design language and all that kind of stuff, and so you need somewhere to …”
Lauritsen: Without product taste, unguided AI outputs converge to generic grayness
“I think if you don't have taste, then you let AI slop converge to sort of grayness, and everything looks the same, and everything's just like, I think you need to have taste to have sort of An opinion and it stands in the world. This is who we are. This is wha…”
Lauritsen: Vibe coding clones hit 90% easily, but enterprise edge cases are hard
“It's very quick to get to the 90% where it looks the same, and in, like, 80% of the cases it works similarly. It's the other 90% that are difficult. You know, it's like ensuring all the edge cases work, and all the unhappy paths, and all the audit locking, and…”
Jacob Lauritsen: AI speed, product speed, and human adoption speed are distinct
“We have this notion internally that there's the speed of AI, there's the speed of our product, and then there's the speed of humans, you know, and they're like, they're not the same necessarily.”
Lauritzen: Internal vibe coding will become the future for enterprise software
“It will be the future. I don't know when, but it will be the future.”
Legora employee vibe coded a custom immigration app in one day
“Ryan, who joined from Canada, we have a team of people joining from Canada, they're all moving to Sweden, and he vibe coded an app to help everyone migrate. So like very specifically, if you're Canadian, these are like all the laws and all the steps you take, …”
Lauritsen: Custom software development makes sense for shallow, highly customized apps
“If it's a shallow app, and it requires a lot of customization from you, maybe you just build it. That's probably actually the right thing to do. If it's a very deep one, there's just too much stuff for you to build, and it's not viable for you to do.”
Lauritsen: PMs should vibe code prototypes to reduce handover costs
“So it's good if PMs do some amount of vibe coding to, like, show very high fidelity, here's a prototype, this is exactly what it looks like.”
Lauritsen: Enterprise Customers Do Not Buy Legora Based on AI Models
“Let's say you took away a model from LIGORA, people would still pick LIGORA. So they don't buy it based on the model.”
Lauritsen: Lawyers Prefer Superior Output Quality Over Fast Latency
“Almost always performance. Performance is more important. Almost always. If you're a lawyer, you can wait two seconds more for the output if it's better. You can probably wait an hour more for the output if it's better.”
Lauritzen: Enterprise IT will evolve to build internal AI tools in 5 years
“I think the internal AI systems role thingy for enterprises, I think IT will, can have like a flowering moment here and, you know, go from being internal IT that's like setting up your computers and whatever to maybe have a system that's building a tons of int…”
Lauritsen: Legora may not need forward-deployed engineers in five years
“We currently do, yeah. But I think that's just, that's an education thing. In five years, maybe we don't at all, to be honest.”
Lauritsen: Harvey was more aggressive with hiring than Legora
“I think they've been more aggressive with hiring, which I actually think I have not been aggressive enough with hiring because I've always tried to have a very, very small, very lean team, which I believed a lot in.”
Lauritsen: Delaying Legora's Developer Experience team was a management mistake
“We have a developer experience team. Relatively new. Again, a mistake I made. I should have staffed that earlier.”
Lauritsen: Legora engineers run 10 background coding agents concurrently
“We have our own background coding agent that they built. That allows each engineer to have, like, 10 different agents running concurrently with, like, all of our local development, a browser, all the iteration stuff.”
Lauritsen: US engineers are less risk-averse while European engineers are more loyal
“In the US, people are generally less risk averse, so like, they'll, they'll be ready to jump on a lot of things to test. I think in Europe, people are more risk averse. People in Europe, and I'm super generalizing now, but I think Generally are more, I don't w…”
Lauritsen: European candidates struggle to properly value tech startup equity
“It's actually something that we had to sort of educate people on in Sweden. And in Europe, I think it's, people are just not used to the venture thing. They don't know how to value equity the same way.”
Lauritsen: Evaluating employees using AI token usage leaderboards is stupid
“Having a leaderboard, a lot of people say this, get a leaderboard and bring up token usage at performance reviews, And that leads to token maxing, which is people just burn tokens just to look good. That's a really stupid way to do anything.”
Stebbings: Cognition and Factory will thrive due to model independence
“I respectfully disagree, and that's why I actually think both Cognition and Factory will do very well, because they're model independent”
Lauritsen: Compute deficits force AI coding tools into harmful vertical integration
“Obviously they don't have enough compute. They can't train their own models. They probably have to train their own models. It's I just think it's a shame that the industry is, is vertically integrated in that way.”
Lauritsen: Traditional IDEs will die and reading lines of code will disappear
“The current shape of an ID will die. Yes. I think we, I don't know what the new, the next ideas, but it's not reading lines of code. It's maybe it's graphical, honestly. Like maybe it's the systems, the architecture that you look at and you review and you plan…”
Lauritzen: Legora now designs all software systems to scale to 100x usage
“Now, now I make sure everything we build will scale to 100 X the usage. I used to say 10 X, but then that was not enough, so now everything needs to scale to a hundred X.”
Lauritzen: System fair queuing should prioritize small interactive requests over bulk tasks
“If you're running a 100,000 cells, you're probably okay waiting a bit. You can go grab a coffee and that's fine, but if you run 10 cells at the same time, those should be really fast.”
Jacob Lauritsen: Elite engineers create more value than superior AI models
“Engineers. For sure. Because the models they change all the time, they get better all the time, but if you have really good engineers, you can build a system that exponentially improves, and that's worth a lot more.”
Lauritzen rejects idea that executives are suited only for specific startup stages
“No, I don't buy that. I think It's, to me, it's a question of how quickly can I solve problems? Am I the person that can solve the problems that we have right now the fastest, or can someone else solve them faster than me?”
Lauritsen: Legora rarely discusses job titles when hiring engineers
“And I only, most of the people that we hire, we don't even talk about their title. We talk about the difficult problems that they're going to work on.”
Lauritzen: Co-locating PMs, designers, and engineers eliminates handoff friction
“If you have a PM and a designer and an engineer, and they just sit together, You can almost not even have the handover. You can just be like, run at this problem and do it together, the three of you know, next week, and then it's done. But if you have them sil…”
Lauritzen: Legora is avoiding hiring remote engineers to prioritize in-person collaboration
“Well, we're very opinionated about who we are and who we aren't. We, if you're a great engineer and you want to be remote, then you probably also want to work on very isolated problems, and that's, that can be fine. We'll, we'll, we probably have those, and we…”
Lauritzen Projects Legora Will Reach 270 Engineers by End of 2027
“If I have to put my, I want to say the lower number let's say two 70.”
Lauritzen: Hiring Untrusted Players Forces Top Engineering Talent to Leave
“I'd rather miss my number and have A players than hit it with B players, because I think as soon as you introduce players, as soon as you have people that you don't trust or that the team doesn't trust, the A players won't stick around.”
Lauritzen: Legora rebuilds acqui-hired code directly into main codebase
“If they'worked on adjacent things or think it's in a similar field or even, you know, unrelated but similar technology, we'll take all their learnings and we might rebuild it into Lagora. I think that's what happens in most cases, but they become, you know, fu…”
Lauritzen: Integrating acquired engineers is easy if they lack ego
“It's surprisingly easy if you hire people with low ego. That's the thing. If you get five great engineers that don't care about their titles or where they sit in the org chart that just want to solve problems, it's surprisingly easy to integrate.”
Lauritzen: Legora Gives Firing Ultimatums to Underperformers Within Two Weeks
“I give really strong feedback after two weeks. What does really strong feedback mean, Jay? Really strong means you're not gonna stay if you don't change this. And feedback is feedback.”
Lauritzen: No Legora Employee Has Recovered From Ultimatum Feedback
“No. But they need to get the chance. And if they do, they stay.”
Lauritsen: Technical engineering managers who have scaled are extremely hard to hire
“I think senior management is extremely difficult to hire for. No yeah, or like engineering directors, engineering managers within, yeah, within product and engineering, yeah. Maybe that's always been difficult, but I think it's because we only have really tech…”
Lauritsen: Legora engineering managers spend most of their time coding
“The way that we have engineering teams is relatively small teams, let's say six people, a PM and an engineering manager. The engineering manager is super technical, you know, spends most of their time coding.”
Lauritsen: Lawyers will shift from contract drafting to high-level risk strategy
“Eventually, lawyers will not be nitty gritty about the language of the contracts. They will work a level above, which is maybe like, What's our negotiation stance? Like what risks are we okay? Which ones are we not okay taking? And not sit and type into word.”