How Bito AI is assisting developers to review codes better

Code review is one of the most overlooked parts of the modern software development lifecycle. It creates a critical bottleneck before any new software updates can ship. Normally, this essential but tedious task of checking teammates' code for bugs and improvements falls on the shoulders of human developers, but serial entrepreneur Amar Goel believes Bito AI can offload that burden onto an AI agent that spots issues and makes fixes on its own.

The Harvard alum who previously took PubMatic, an advertising technology firm, public in 2020 after growing it from a 2006 startup, is now betting that AI can automate software development's most manual process. 

“We believe code review is a super critical part of the developer lifecycle. It’s becoming more critical, because as AI writes more code, developers will have to review more code,” Goel, CEO of Bito AI, tells YourStory. 

“Code review may become a majority of the job the developers are doing… AI agents will write a bunch of code, and they will need to make sure it’s doing the right thing, designed properly, and integrating well with existing software,” he adds. 

Founded by Goel, Anand Das, and Mukesh Agarwal in 2021, the AI developer platform assists developers in reviewing and improving code by deeply understanding their codebase.

Product offerings

The Menlo Park, US-based firm’s proposition is AI that handles the grunt work of code review so human programmers can focus on building new features. 

Bito AI provides line-by-line feedback across more than 50 programming languages and integrates directly into existing workflows like GitHub and Bitbucket, where developers store and manage their code projects.

“Initially, we explored generative AI to see how it could support software developers. We had launched a chat product for them, but soon realised AI could play a much deeper role in assisting across the entire development process,” Goel says.

“Everyone’s been very focused on generating code with AI, and I think that’s super critical. But there are a lot of other things developers need to do to be successful, and that organisations and teams need to do to get that code into production, get it live, and ensure its quality code,” he adds. 

One of its core offerings includes AI Code Review Agent, its flagship product. This helps developers to review pull requests by looking at code changes and file relationships. 

It integrates directly into GitHub, GitLab, and Bitbucket workflows to provide developers with line-by-line feedback, issue detection, performance optimisations, and one-click fix suggestions across several programming languages.

It also gives clear feedback on code quality, points out bad practices, and highlights any potential security issues, with all suggestions shared as comments. 

The platform also offers Bito Wingman, a coding agent capable of taking meaningful action across the development ecosystem. Developers can interact with it to execute complicated commands and take action. 

For example, Bito has reviewed more than 184,500 lines of code for US-based Gainsight, integrating into GitHub to automate reviews and speed up decisions. It further identified over 2,600 issues, including duplicated code, memory leaks, and inconsistent patterns.

Road ahead

In May, Bito raised $5.7 million in a seed extension round, led by Vela Partners, with participation from NextView Ventures, Maxitech Ventures, Eniac Ventures and others. 

The company has so far raised $8.8 million in total. 

“Developers used to write 200 lines of code a day. Now AI is going to write 10,000 lines of code a day. Now you've got to review that code, test that code, figure out how to get it to production with all the same manual tools you had before. The entire software development lifecycle chain is going to change. That's where we've been really focused on at Bito and really excited about at Bito is how we help developers rework their processes in this new GenAI,” Goel explains. 

However, the serial entrepreneur believes that the platform entirely catches every single potential issue in code, and still advocates having a human reviewer take a look at the code and double check for potential errors. 

“But what we are finding is that pull requests are closing 90% faster. It’s happening in one-tenth of the time. What used to take me maybe four days and 90 hours from asking for a code review to then merging that code; now it's happening in nine hours. The open-to-merge time is dramatically faster, which means that teams are able to move much faster,” he adds. 

The platform claims to process over 10,000 code reviews weekly for enterprises such as Gainsight and SaaS major Whatfix, delivering 89% faster software updates and generating $14 in value for every $1 spent.

Currently, the company follows a per-developer, per-seat licensing model and is targeting mid-market enterprises with engineering teams of 50 to 1,000 people.

“Big enterprises have their own processes and workflows. They want the agent to work their way — they don’t want a cookie-cutter solution. It’s about working with something that can customise to their workflow versus an SMB solution that has a couple of toggles and is easy to set up, but doesn’t give the richness of review they need,” he says.