Pull Request/Merge Request Summaries & Reviewer

From use case: Pull Request/Merge Request Summaries & Reviewer

Visma, a Norwegian provider of business software, deployed CodeRabbit to improve the process of handling an average of 600 pull review requests related to a central application repository of more than 3 million lines of code, some of it more than 20 years old. Seref Boyer, chief architect of Visma’s R&D unit says CodeRabbit helps developers find errors that manual reviewer miss, including typo errors, null pointers, and static code, flags issues in untouched legacy code: “It’s good because then you are also aware of other things that can be done to improve the code and reduce technical debt,” Boyer says, according to a case study by CodeRabbit.

Microsoft has built an AI-powered code review assistant to augment pull request reviews. What began as an internal experiment, now supports over 90% of the company’s 600,000 pull requests per month, according to a 2025 blog post by Sneha Tuli, a principal product manager at Microsoft. “It helps our engineers catch issues faster, complete PRs sooner, and enforce consistent best practices—all within our standard development workflow,” Tuli said. “AI reviews the code changes and leaves comments just like a human reviewer would. These comments show up in the PR discussion thread, so the author and other reviewers see them and can act (just as if a colleague had made the remarks).” The tool suggests changes that the author can accept with a click, with changes “attributed to the commit history, preserving accountability and transparency,” Tuli said.

Return on investment analysis highlights the importance of a proper implementation strategy. AI-generated summaries significantly reduce the time reviewers spend on initial tasks, allowing them to allocate more time to thorough analysis. Success factors include starting with small pilot teams, establishing clear guidelines for when to rely on AI, and maintaining feedback loops to continuously improve model performance. Engineering leaders note that AI tools elevate the entire code review discussion rather than just saving time.