Amar Goel is the CEO of Bito, which builds AI agents for developers centered on deep codebase understanding. Its first agent is an AI code reviewer for GitHub, GitLab and Bitbucket that also runs inside IDEs. Amar is based in Menlo Park, and Sam noted he is an Amazon alum. He joined us in Episode 23 (December 5, 2025).
Amar's pitch starts with where AI coding tools fail (06:21). Everyone uses Cursor, Claude Code, OpenAI Codex and Copilot, and they work almost one-shot on new projects. Legacy codebases are different.
But when I start to get into my existing kinda gnarly code base that's been built for five, ten, twenty years, and I have millions of lines of code and hundreds of repos and different services, all of a sudden, tools start to kind of break down.
— Amar Goel, Episode 23
Bito indexes the whole codebase, building symbol indexes and abstract syntax trees, so its review sees how services connect. Amar's example: it can flag that a change to one service will break an upstream service that calls it. Teams also run Bito's review inside Cursor, Windsurf, Claude Code, VS Code and JetBrains. He was candid that a simple vibe-coded project probably doesn't need it; an enterprise does.
And also when a bug gets into production, the developer a lot of times didn't even write the code. So nobody really knows what's going on anymore.
— Amar Goel, Episode 23
So a PR that used to take fifty hours to merge is now merging in five to seven hours because most enterprise teams still have one or two people who manually go and review a PR.
— Amar Goel, Episode 23
Amar also mentioned that Bito uses billions of tokens a day and sometimes hits model providers' capacity limits. He teased a launch for the week after the episode that would take Bito's codebase understanding beyond code review, but did not share details.
Bito didn't have a formal Slack bot yet, so Jordan opened the show (01:34) with a Bito Slack bot prototype built in Google AI Studio from a basic prompt. It imitates the Slack interface. "Bito help" lists the commands, "Bito last" shows a review of his last PR (a dark mode toggle that failed on a critical security bug), and "fun," "surprise" and "disco" return a developer joke, a haiku and a dancing disco mode.
Amar joked that it should go to production right away; Jordan said it was not ready. Amar added that customers have asked for exactly this kind of Slack notification when a review finishes or finds issues. More on the category on our AI in developer tools page.
Amar Goel is the CEO of Bito, based in Menlo Park in the Bay Area. Sam noted on the show that he is an Amazon alum. He joined Built This Week in Episode 23 in December 2025.
Bito builds AI agents for developers centered on deep codebase understanding. Its first agent does code review in GitHub, GitLab and Bitbucket, and teams can also run it inside IDEs like Cursor, Claude Code, VS Code and JetBrains.
Amar says tools like Cursor and Claude Code work well on new projects but break down in legacy codebases with millions of lines, hundreds of repos and many services. Bito builds symbol indexes and abstract syntax trees across the codebase so it can catch problems such as a change that breaks an upstream service.
Amar said PRs merge about 10x faster on average: a PR that took fifty hours now merges in five to seven hours, because Bito's feedback arrives in about five minutes. Customers also see about 55% fewer reverted PRs in the following month.
Yes. Teams can choose an essential mode for production-level issues or more comprehensive feedback, add custom rules, and let the agent learn from what the team actually acts on.
