Lubeno

A fresh new take on pull-request iteration for the era of AI-generated code.

Role
Co-founder · Product Design · Frontend
Timeline
2025 – Present
Platform
Web application
Status
lubeno.dev
200+ users
Lubeno product screenshot
01

Problem and Context

Pull-request workflows haven't evolved to match how code is written today. Developers use AI-assisted coding more often than not, and the DevOps market keeps pouring resources into code generation efficiency. However, code review tools haven't seen a significant change in 20 years. The human reviewer is still the one accountable, now for a larger volume of code they didn't write themselves. There is a high cost to this. Developers are frustrated, code quality suffers, and pull requests sit waiting while the work built on top of them can't move forward.

78%
say developers are writing and committing code faster since adopting AI
+91%
longer pull-request review time on teams with high AI adoption
43%
of developers trust the accuracy of AI output — the rest are unsure or skeptical
02

Research and Discovery

We started from our own lived problem as developers working in teams, and then decided to validate it. We followed discussions on public developer forums, published articles that provoke reactions and pull feedback, and finally started direct conversations with real users. Each stage was validated enough to justify the next.

1 · The problem
My own daily frustration as hypothesis
2 · Forum research
Hacker News, X and Reddit on current tools
3 · Publishing
Articles that pull feedback from developers
4 · Direct feedback
Individual conversations with our users

"What I really want when I'm reading source code is to hover over a line and ask 'why was this done this way?' That reasoning almost never makes it into the docs but it's exactly what lives in the review discussion."

— Developer, from our user conversations
03

Design Decisions

Our review flow is built around three decisions that remove the waiting and the back-and-forth discussions that slow down the process:

Stacked pull requests
PRs build on top of each other, so users can keep working on new ones while previous ones are under review.
Code evolution tracking
The tool intelligently tracks what each reviewer has already seen, and comments never get lost in the noise of new changes. Reviewers can focus on what changed since their last review.
Git-compatible, Jujutsu-native
We offer backwards compatibility with Git, while Jujutsu users get its full velocity benefits through a visual tool.

"In every review tool I've used, I never know what order to review files in. As the author, I just want to float the important files to the top so reviewers look there first."

— Another developer we spoke with
Stacked PRs
How stacked PRs look
Code review flow
The code review flow
Key pivot: after testing the first prototype, we realized users needed a more linear flow rather than the branching model we initially designed.
04

Building It

My role across the stack

I worked on the research, visual design and frontend implementation of Lubeno. Because I build what I design, the interface arrived grounded in what the hybrid Git/Jujutsu architecture could actually support, and design decisions fed directly back into how we shaped the product.

Designed for gradual adoption

Organizations that host and collaborate on Git can adopt Lubeno gradually since there's no migration cliff. A free version serves individual developers and small teams, and a comprehensive paid subscription serves larger companies.

05

Outcome

200+
users hosting code with us
2025
from idea to a live, growing product
100%
hosted in Europe

What I learned

  • Design for emerging workflows rather than optimizing existing ones.
  • Knowing when to pivot on user feedback versus staying the course on the original vision.

What's next

  • Continuous Integration, protection rules and richer approval options.
  • IDE integrations and an AI agent that summarizes and pre-reviews pull requests with user-set rules.
Next case study
Restaurant Mobile App La Fermina