As Silicon Valley races toward a early wherever AI agents do astir of nan package programming, a caller problem is created: uncovering nan AI-generated bugs earlier they are put into production. Even OpenAI is dealing pinch specified issues, a erstwhile worker has described.
Newly funded startup PlayerZero has created a solution: usage AI agents trained to find and hole problems earlier nan codification is put into production, nan startup’s CEO and sole founder, Animesh Koratana, tells TechCrunch.
Koratana created PlayerZero while he was astatine nan Stanford DAWN laboratory for instrumentality learning nether his advisor and laboratory founder, Matei Zaharia. Zaharia is, of course, a famed developer and nan co-founder of Databricks; he created its foundational exertion while moving connected his ain doctorate.
PlayerZero connected Wednesday announced that it raised a $15 cardinal Series A led by Foundation Capital’s Ashu Garg, an early Databricks backer. This follows a $5 cardinal seed led by Green Bay Ventures and respective noteworthy angels including Zaharia, Dropbox CEO Drew Houston, Figma CEO Dylan Field, and Vercel CEO Guillermo Rauch.
During his clip astatine Stanford DAWN, Koratana, now 26, was moving connected AI exemplary compression exertion and “got exposed to connection models really early on,” he says. He met nan developers who crafted immoderate of nan first AI coding assistance tools.
It deed him past that “there’s this world successful which computers are going to constitute nan code. It’s not going to beryllium humans anymore,” Koratana told TechCrunch. ”What’s nan world going to look for illustration astatine that point?”
He knew earlier nan word “AI slop” was moreover coined that these agents were going to nutrient codification that collapsed things conscionable arsenic their quality overseers do.
Techcrunch event
San Francisco | October 27-29, 2025
That problem would besides beryllium exacerbated by truthful galore agents cranking retired truthful overmuch much codification than has ever been written before. It won’t ever beryllium applicable for humans to cheque each AI-written codification for bugs aliases hallucinations. And nan rumor becomes moreover much aggravated for nan large, analyzable codification bases that enterprises trust upon.
ProjectZero trains models “that really profoundly understand codification bases, and we understand nan measurement they’re built, nan measurement they’re architected,” Koratana says.
His tech studies nan history of an enterprise’s bugs, issues, and solutions. When thing breaks, his merchandise tin past “figure retired why and hole it, and past study from those mistakes to forestall them from ever happening again,” Koratana says. He likens his merchandise to an immune strategy for ample codification bases.
Landing Zaharia, his adviser, arsenic an angel was a first measurement to fundraising, but nan infinitesimal that really validated his thought was erstwhile he showed a demo to different celebrated developer: Rauch. Rauch is laminitis of triple unicorn developer instrumentality company Vercel and creator of nan celebrated open-source JavaScript model Next.js.
Rauch watched Koratana’s demo pinch liking but skepticism, asking really overmuch of it was “real.” Koratana replied that this was codification “running successful production. Like, this is simply a existent instance. And he was quiet,” Koratana says. Then his soon-to-be-angel investor responded, “If you tin really lick this nan measurement that you’re imagining, it’s a really large deal.”
Of course, PlayerZero isn’t unsocial successful attempting to lick nan AI-generated bug problem. Just past week, Anysphere’s Cursor launched Bugbot to observe coding errors, arsenic conscionable 1 example.
Still, PlayerZero is already gaining traction for its accent connected ample codebases. While it was conceived for a world wherever agents are nan coders, it is presently being utilized by respective ample enterprises that usage coding co-pilots. For instance, subscription billing institution Zuora is 1 of nan startup’s marquee customers. Zuora is utilizing nan tech crossed its engineering teams, including to watchdog its astir precious code, its billing systems, it said.