How freee saved months of reviewer time with CodeRabbit

Why freee loves CodeRabbit: Stronger review quality, better usability & easier budgeting

CodeRabbit first came to freee through personal use. Software Engineer Jeong Jaesoon had used the free version for solo development and saw that it offered something missing from other review tools already in the mix internally.

“The other tool that was already implemented only presented reviews one-sidedly, and I felt it was lacking in the inability to have dialogue or properly configure settings like ‘I want you to review from this perspective,’” Jeong said. “Through using CodeRabbit in personal development, I realized that the usability in those areas was very good, and I thought it would match our internal challenges, so I proposed a comparative verification.”

When freee compared CodeRabbit with alternatives, review accuracy stood out as a major differentiator in favor of CodeRabbit. freee also valued CodeRabbit’s operational fit. The company appreciated the ability to add and manage custom coding rules through Learnings, as well as the dashboard and chat-based workflows.

High acceptance rate = Devs find things they wouldn’t otherwise

That accuracy shows up not just anecdotally, but in the results freee saw in practice. Across freee’s use of CodeRabbit, the platform achieved an overall acceptance rate of 48.5%, meaning nearly half of CodeRabbit’s suggestions were accepted by developers.

The impact also cut across multiple review categories. Acceptance rates reached:

• 52.3% for functional correctness
• 51.1% for data integrity and integration
• 49.8% for performance and scalability
• 49.1% for stability and availability
• 45.4% for security and privacy
• 44.8% for maintainability and code quality.

Together, those numbers suggest that CodeRabbit was not only catching superficial issues, but contributing across the range of concerns engineering teams have to manage in production systems.

Flat-rate pricing allowed for easier budgeting

Unlike usage-based tools, CodeRabbit’s flat-rate subscription model made budgeting far easier.

“With infrastructure like AWS, if usage spikes, you can explain it as ‘access increased and revenue went up.’ However, with development tools, usage fees don't necessarily correlate with productivity. For development tool budgeting where flat-rate pricing was traditionally the norm, predictable costs were a very significant reassurance,” Yuhei said.

freee began with about 30 users, tested CodeRabbit for roughly a month, and then expanded the rollout to 570 seats, to “everyone who writes code,” Yuhei said. Today, the platform is active across 285 repositories, giving the company broad review coverage across its engineering environment.

Results: Weeks of reviewer time saved & devs love it

32.8 weeks of reviewer time saved

At scale, the benefits of CodeRabbit have been tangible. CodeRabbit has saved reviewers the equivalent of 32.8 weeks of reviewer time in the last six months, helping teams absorb higher pull request volume without asking humans to shoulder the full burden alone. That matters because freee does not see AI review as a replacement for human judgment. Instead, the company uses CodeRabbit to create a better division of labor between AI and people. Developers are encouraged to work through CodeRabbit’s suggestions before requesting human review, allowing AI to catch basic issues first and freeing human reviewers to focus on higher-order concerns like architecture, intent, and design.

84% satisfaction = Devs overwhelmingly love CodeRabbit

On an internal survey, the average satisfaction score was four out of five, about 84% of freee’s engineers said they were satisfied with CodeRabbit and roughly 70% said CodeRabbit had reduced their review burden. For junior developers, that means greater confidence before handing off work for review. For senior developers, it means less time spent on routine comments and more time spent on the issues that really require human judgment. “For junior members, there's a sense of security that ‘I can self-check before sending a review request,’ and for senior members, they can trust that ‘AI has already completed basic checks,’ creating positive effects for both sides,” Jeong said.

Added features that make reviews faster

freee also highlighted CodeRabbit’s walkthroughs and summaries as especially helpful in understanding pull requests faster. And there was another benefit that mattered in everyday team life. CodeRabbit helped remove some of the interpersonal friction around small corrections that can be awkward for humans to point out repeatedly.

For freee, adopting AI-assisted development increased output, but it also exposed the limits of a review process that depended too heavily on human bandwidth alone. CodeRabbit helped the company respond by improving review quality, reducing burden on engineers, and saving significant reviewer time across a large development footprint.

More broadly, freee sees AI review as essential infrastructure for modern software development. “I think it's one of the absolutely necessary parts in the AI coding era,” Jeong said of CodeRabbit. “Both having AI create things and having AI do the work is important, but from the perspective of how to ensure quality, I think we can't consider it without review AI as a set, or it becomes a bottleneck.”

With active use across 285 repositories, nearly half of suggestions accepted, and the equivalent of 32.8 weeks of reviewer time saved, CodeRabbit has become an important part of how freee keeps development moving without letting code review slow it down.