When Harjot Gill sold his first startup, Netsil, to Nutanix in 2018, he didn’t just cash out; he continued to observe how software was being built. Years later, while running his next venture, FluxNinja, he spotted something that would shape his next big move, and eventually lead to CodeRabbit developers quietly adopting AI coding assistants, such as GitHub Copilot.
The promise was speed, but Gill saw the catch. “It was very clear to me that as a second-order effect, it is going to cause bottlenecks in the code review,” he told TechCrunch. In other words, AI could churn out code quickly, but much of that code was flawed, and humans had to spend hours combing through it.
That observation became the spark for CodeRabbit, an AI-powered code review platform he launched in early 2023 (and later merged with FluxNinja).
The bet paid off. Developers today are leaning heavily on AI to write code, but they are also drowning in fixes. CodeRabbit has stepped in as a digital coworker that learns a company’s codebase, flags bugs, and provides feedback that can cut the human review workload by half. The startup already works with big names like Chegg, Groupon, and Mercury, plus more than 8,000 other businesses.
Related: Warp Launches “Warp Code” to Make AI Coding Agents More Transparent
And the growth? Hard to ignore. CodeRabbit says it is pulling in over $15 million in annual recurring revenue, growing 20% every month. Investors have noticed too. On Tuesday, the company announced a $60 million Series B round led by Scale Venture Partners, with backing from Nvidia’s venture arm NVentures, CRV, and others. The raise values CodeRabbit at $550 million and brings its total funding to $88 million.
Of course, the AI review space is heating up. Competitors like Graphite (which raised $52 million earlier this year) and Greptile (reportedly in talks for $30 million) are pushing hard. Even big AI players like Anthropic and Cursor are bundling review features into their coding assistants.
But Gill believes the future belongs to standalone specialists. “CodeRabbit is a lot more comprehensive in terms of depth and technical breadth than bundled solutions,” he argued. Thousands of developers seem to agree, they are already paying $30 a month to let the Rabbit run its checks.
Here is the bigger truth: AI can write code, but it still cannot be fully trusted to clean up its own mess. That gap has given rise to a curious new role in engineering teams, the “AI code cleanup specialist.” And until AI can review itself without help, startups like CodeRabbit are set to thrive.
If AI is already coding and reviewing, will humans end up managing code, or managing the AI that manages the code?
