CodeRabbit raises $16 million to use AI to code reviews

 Code review the peer review of code that helps developers improve code quality is time-consuming. According to one source, 50% of companies spend two to five hours per week. Without enough people, code reviews can be overwhelming and distract developers from other important work.

Harriot Gill believes that the use of artificial intelligence can greatly automate code reviews. He is the co-founder and CEO of CodeRabbit, a company that uses artificial intelligence models to analyze code to provide feedback. Before founding CodeRabbit, Gil was Sr. CTO at Nutanix, a data center software company. He joined Nutanix in March 2018 when it acquired his startup Netsil. Gill said the CodeRabbit platform uses "advanced AI reasoning" to automate code reviews to "understand the intent of the code" and provide developers with "actionable" and "human" feedback.

“Traditional static analysis tools and curves are rule-based and often produce high false-positive rates, while peer review is time-consuming and subjective,” Gill told. "By comparison, CodeRabbit is an AI-first platform."

These are bold statements with lots of buzzwords. Unfortunately, CodeRabbit's anecdotal evidence suggests that AI-driven code reviews tend to be inferior to human code reviews. In a blog post, Graphite's Greg Foster talks about internal experiments using OpenAI for GPT-4 code review. Although the model can capture useful information (such as small logic errors and typos), it can generate a large number of false positives. Even attempts at fine-tuning have not significantly reduced these problems, Foster said.

These are not revelations. A recent Stanford University study found that engineers who use code generation systems are more likely to introduce security vulnerabilities into the applications they develop. Copyright is also an ongoing issue. There are also logistical drawbacks to using artificial intelligence for code review. As Foster points out, more traditional code reviews force engineers to learn through meetings and conversations with fellow developers. Offloading reviews threatens this knowledge sharing.

Gill feels differently. “CodeRabbit’s AI-first approach improves code quality and significantly reduces the manual effort required in the code review process,” he said. Some folks are buying the sales pitch. Gill claimed that about 600 organizations currently pay for CodeRabbit's services and that CodeRabbit is running pilots with "several" Fortune 500 companies.

It also has investments: CodeRabbit today announced a $16 million Series A round led by CRV, with participation from Flex Capital and Engineering Capital. Bringing the company's total to just under $20 million, the new money will be used to expand CodeRabbit's 10-person sales and marketing functions and product offering, with a focus on improving its security vulnerability analysis capabilities.

"We will invest in deeper integrations with platforms like Jira and Slack, as well as AI-powered analytics and reporting tools," Gill said, adding that Bay Area-based CodeRabbbit is in the process of establishing a new office in Bangalore as it doubles approximately the size of the team. "The platform will also introduce advanced artificial intelligence automation for dependency management, code refactoring, unit test generation, and documentation generation."