When global firms like Goodwin Procter ($2.4B revenue) entrust your software with their most sensitive legal work, you can’t afford bugs.

Legora faced this dual challenge: scaling from 15 to 35 engineers while serving 250+ law firms across 20 countries, with zero tolerance for defects.

That's where cubic came in.

Hypergrowth under zero‑bug constraints

"We're 160 people now. It's a hypergrowth startup in the truest sense of the word."

— Jacob Lauritzen, CTO at Legora

When Jacob joined, Legora was building “Cursor for lawyers”: AI tools that meet enterprise legal standards, where a single bug can mean millions in liability. Code review had become the bottleneck, so Jacob took a pragmatic approach.

Comparing cubic versus 3 other tools

"Let's try out four at the same time. Everyone got bombarded with options, and then I did a survey: which one is the most helpful? cubic was clearly the winner: it understood our codebase better than the others."

— Jacob Lauritzen

Developers tested four tools in parallel for a week and voted.

cubic stood out not just for bug‑catching, but for how easily custom rules could encode Legora’s conventions. Example: it flagged service functions returning raw promises instead of Legora’s standard async result pattern.

The difference was clear: cubic understood their conventions, patterns, and context. Not just finding bugs, but understanding how Legora writes code.

The Redis cache bug cubic caught

The validation came when Jacob was rewriting their Redis client.

The code looked perfect, but cubic caught something subtle.

"cubic actually flagged that I was purging the cache for all of these errors, but not for the initial connect. Humans look at code flow and architecture. But some of those subtle bugs are just really difficult even for people to notice."

— Jacob Lauritzen

The connection failure wouldn't purge the cache, while other errors would. An inconsistency so subtle that human reviewers could easily miss it.

From bottlenecks to guardrails

Today, cubic catches what humans miss: bugs, convention violations, security issues, and performance problems.

"We're going to enforce that when you do a PR, you first have to address the AI’s comments and the self-review. Then you request another human’s review."

— Jacob Lauritzen

cubic also enforces Legora’s testing requirements from project READMEs, flagging uncovered edge cases and off‑pattern tests.

By the time humans review, the obvious issues are gone. Reviewers focus on architecture and mentorship.

Impact: 1,001 fixes, 31% faster merges, 10‑hour average merge time

The numbers tell the story:

• 1,001 issues fixed thanks to cubic

• 31% faster PR merges since installing cubic

• 10-hour average merge time: exceptional for a 35-person engineering team

Start your free trial and ship faster without sacrificing quality.