Harjot Gill led Fluxninja, an statement startup, which he founded a few years after selling his first startup Netsil to Nutanix In 2018, when he noticed an interesting trend.
“We had a team of distant engineers who began to generate AI code to Github Copilot,” said Gill. “We saw that adoption was happening and it was very clear to me that as a second row result, it would cause bottlenecks in the Code Review.”
At the starting of 2023, Gill began Codrabbit, the AI -powered code review platform, and purchased Flexninja.
Gill’s prediction got here true: programmers repeatedly use AI coding assistants to generate code, but the power is often employee, forcing engineers to spend a lot of time on corrections.
Codebbit may also help catch some errors. According to Gill, the activity increases by 20% monthly and currently earns over $ 15 million of annual repetitive revenues (ARR).
Investors imagine that the startup increase is exciting. On Tuesday, Codrabbit announced that he collected $ 60 million B series, valuing the company at $ 550 million. The round, which led to the complete financing of the startup as much as $ 88 million, was run by SCLE Venture Partners with the participation of Nventures, the Venture Capital Nvidia arm and returning investors, including CRV.
Codebbit helps corporations equivalent to Chegg, Groupon and Mercury, in addition to over 8,000 other corporations, save time on the famous frustrating task of the code review, which has turn out to be much more time consuming with the development of the code generated by AI.
TechCrunch event
San Francisco
|.
October 27-29 2025
Because Codrabbit understands the company’s code database, he can discover errors and provide feedback, behaving like a collaborator, said Gill. He added that corporations using Codrabbit can reduce the number of people working on the code review by half.
As in the case of most AI areas, Codrabbit has competition. The startup rivals include graphite, which provided a B -series with a value of $ 52 million run by Accel at the starting of this 12 months, and Greptile, which, as we reported, talks about a round of $ 30 million from Benchmark.
(*60*) the leading assistants encoding AI, equivalent to the Anthropic code and cursor, also offer options for reviewing the code powered by AI, Gill betting that customers will in the long term prefer to supply the offer in the long term. “Codrabbit is much more comprehensive in terms of the depth and technical width than packet solutions,” he said.
Whether his forecast proves to be correct. But for now, 1000’s of programmers will likely be glad to pay Codrabbit 30 USD monthly.
Even with the growing popularity of tools for reviewing AI code, equivalent to Codrabbit, AI solutions, still can’t be fully trusted to repair errors and “useless” code written by AI. The unreliability of the code generated by AI gave the creation of a latest corporate role: Code cleansing specialist.