How Visma enhanced code quality and streamlined reviews
Catches hard to find bugs
Faster code reviews
Great PR context & summaries
More secure merges
Overview
For Seref Boyer, the Chief Architect of a Visma R&D unit, managing their large codebase is high stakes since millions of users rely on the applications his company powers. Visma, a major player in the European ERP market, develops internal solutions for finance, logistics, warehousing, and project accounting across Northern Europe. That makes quality assurance a mission-critical focus for Visma.
Seref's team is made up of approximately 80 developers, including 40-50 developers focused on product development and research. Their core product isn't just another web application—it's a powerful ASP.NET system built on the Microsoft .NET framework with a large, legacy codebase that’s over 20 years old.
This complex, decades-old codebase creates a perfect storm of challenges for new developers who often struggle with onboarding and introduce bugs and errors into the codebase in their first months as they get up to speed.
Seref learned about CodeRabbit when another team shared their positive experience integrating the tool on their project. This internal validation convinced Seref to adopt CodeRabbit for his team, as well. The result? Increased error detection and unprecedented clarity in pull requests that transformed his team’s development cycle.
Challenge: Taming a large, legacy codebase
Before CodeRabbit, Visma's development process was complex. Human reviewers carefully examined their team’s 600 monthly pull requests (a process still central to their workflow). Still, even the most eagle-eyed developers faced limitations when trying to protect a legacy codebase of this magnitude:
- Navigating legacy complexity and cross-dependencies: \ The application's central repository contains more than 3 million lines of mostly legacy code dating back to the early 2000s. This extensive, interlinked system meant that even minor updates could cause unforeseen ripple effects. "If you make changes, they could affect other parts as well if you are unaware of them," Seref emphasized. “There are several cross-dependencies where packages interact with each other.” The codebase's sheer size made understanding these relationships difficult for any one developer.
- Accelerating onboarding and mitigating risk for new devs: \ For newcomers, this complexity created a Mount Everest-sized learning curve. "The code is complex for new developers who are starting to work with it. It takes time to onboard them and they need to understand the risk when they do refactoring or code changes," Seref explained. Given what was at stake with their ERP product, this created a bottleneck for new talent to start contributing effectively and safely.
- Overcoming limitations of manual review in edge cases: \ While human reviewers excelled at understanding business logic, critical technical errors often slipped through even the most careful review. "Even human reviews can miss that if they don't pay attention," Seref acknowledged.
Why Visma loves CodeRabbit
Catches bugs & issues human reviewers miss
One key reason Visma loves CodeRabbit is that it helps sniff out bugs that human reviewers often miss. CodeRabbit’s capabilities identify subtle but critical technical errors that could otherwise slip into production.
"The feedback I get from developers is that they like that it shows them typo errors, null pointers, and static code," Seref shared. He praised CodeRabbit’s ability to discover errors before they reach production, especially in those notoriously difficult scenarios:
It's the edge scenarios that are usually difficult to catch. It's a good tool to get these edge scenarios.
- Seref Boyer, Chief Architect
Additionally, Seref liked CodeRabbit's tendency to flag 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." CodeRabbit, Seref shared, highlighted key opportunities for manually identifying and addressing technical debt within their evolving legacy system.
Revolutionary PR context with AI summaries & sequence diagrams
CodeRabbit's AI-generated summaries and sequence diagrams were an instant hit with Visma's developers, providing clarity that helped them understand complex changes more easily.
CodeRabbit also gives developers an excellent overview with sequence diagrams. Our devs like the sequence diagrams, and I think one of the developers even mentioned that it gives a better summary than they could write by themselves.
- Seref Boyer, Chief Architect
This visualization benefit extends to the dev writing the code, as well. "It gives a nice picture of what you have done,” he said.
Accelerating the code review cycle
While Visma wasn't primarily focused on increasing raw output, they saw a marked improvement in review speed after adopting CodeRabbit.
The code review part has been faster. It has been speeding up the time it takes, which is remarkable. - Seref Boyer, Chief Architect
This acceleration allows developers to move on to other critical tasks more quickly.
Intelligent adaptability
One thing that Seref loves about CodeRabbit is how Visma’s developers can customize their reviews – something they’ve achieved by integrating PATH instructions and using some of CodeRabbit’s other customization features – something that’s critical with a legacy codebase. Seref observed, "Our devs created 24 learnings in the last 30 days. Developers are interested in customizing their settings.”
Results: Protection for their mission-critical legacy codebase
Elevated production readiness & bug prevention Since they implemented CodeRabbit, it has caught several critical bugs during Visma’s pull request stage, significantly reducing the number of incidents they have. "I believe that it is helping us improve our code and product quality,” shared Seref. “That's the goal: To increase code quality and product quality by using an AI tool because it can see errors that are not easy to see for humans," Seref confirmed.
Enhanced context led to streamlined reviews The AI-generated summaries and sequence diagrams provide instant clarity, transforming how developers grasp complex changes and how quickly they can complete their reviews. "It gives a nice picture of what you have done and the changes. Not only for you, the one making the changes, but also for the human who could review it at a later step," he shared. This rapid comprehension significantly reduces back-and-forth communication, resulting in faster reviews overall. "The code review part has been faster," Seref noted.
Freeing human reviewers to focus on business logic By detecting technical errors, CodeRabbit frees Visma's human reviewers to focus their expertise on higher-value tasks, such as validating complex business logic and ensuring architectural alignment with legacy code. This partnership between AI and humans creates a more effective development cycle, according to Seref. He recently explained to a colleague, "I think it's terrific for detecting the local errors. That’s hard to do even in human reviews."
CodeRabbit = More secure merges for their large codebase
Before CodeRabbit
- •Large, complex codebase = more bugs in prod
- •Onboarding devs introduce new bugs
- •Difficulty reviewing edge cases
After CodeRabbit
- More bugs caught thanks to codebase context
- Fewer bugs
- CodeRabbit finds edge cases that humans miss
For Seref and his team at Visma, CodeRabbit isn't just another tool in their development lifecycle – it's become a critical partner in tackling the complexities of large-scale ERP development. "It's very good at detecting localized errors. to give a good picture of the changes which have been applied to the system," Seref explained.
CodeRabbit stands out for Visma’s team because of its unique ability to catch issues that might otherwise slip through the cracks. As Seref puts it, these are problems that humans have difficulty identifying. This added quality check ensures Visma continues to deliver mission-critical software solutions that meet the highest standards of quality and reliability.
The success Seref's team experienced with CodeRabbit has already sparked a wave of interest across the organization, with other Visma teams eagerly joining in as well.
We have at least two other organizations who are buying or adding new licenses because we shared our experience.
- Seref Boyer, Chief Architect
Engineering team size
40-50 developers
Language
ASP.NET, .NET framework, C#, Vue JS, React, JavaScript.
Challenge
Navigating a massive, decades-old codebase with intricate cross-dependencies while ensuring top-tier code quality and overcoming human review limitations.
