Gitar raises $9M to clean up the mess AI-generated code is making
Gitar came out of stealth Wednesday with $9 million in funding and a simple pitch: use AI to fix the problems AI-created code is causing.
The startup, founded by Ali-Reza Adl-Tabatabai (previously Intel Labs, Google, and Uber), sells a platform that deploys AI agents to review code, manage CI workflows, and actually fix issues - not just flag them.
The problem is real.
“Vibe coding” has unleashed a flood of AI-generated code onto engineering teams. A recent report from CodeRabbit found that AI-generated code introduces significant bugs and quality issues that senior engineers then have to clean up before anything ships. This bottleneck has become known as “code overload.”
“Generation produces code; validation makes it trustworthy,” Adl-Tabatabai told TechCrunch. “Gitar is the workflow agent that owns that process, orchestrating reviews, tests, and diagnostics end to end.”
What it actually does.
Gitar plugs into existing GitHub workflows. Its agents review pull requests, analyze CI failures, and can auto-apply fixes. The platform distinguishes between genuine code issues and infrastructure noise. It also deduplicates failures and retries flaky tests automatically.
Customers can also build custom agents for specific security or maintenance tasks.
Who is using it.
The company lists OpenMetadata, SoFi, XFactor.io, and Uber’s Cadence team as customers. A SoFi engineering director said Gitar shortened their mobile CI merge time by identifying failures and providing “AI-powered remediations” that engineers can apply directly from the PR without switching contexts.
An engineering manager at Uber noted that Gitar caught bugs their team missed and could create follow-up issues directly from code changes - tasks that would otherwise require complex GitHub Actions workflows.
The funding.
The $9 million round was led by Venrock with participation from Sierra Ventures. The two-year-old company is betting that as AI code generation accelerates, automated validation will become essential infrastructure rather than a nice-to-have.
They’re probably right. The math is simple: if AI lets junior developers produce 10x more code, you either automate review or you drown.