Playerzero collects $ 15 million to prevent artificial intelligence agents

With the Silicon Valley race towards a future by artificial intelligence agents with most software programming, a new problem is created: the errors created by artificial intelligence before putting them in production. Even Openai is Dealing with such issues, A former employee has been described.

Recent financing startup Playerzero It has created a solution: I used trained artificial intelligence agents to find problems and repair them before placing the code in production, says CEO of Startup and the only founder, Animesh Koratana, Techcrunch.

Corata was created a player as a player while he was in the Stanford Laboratory without automatic learning under his consultant and laboratory founder, Matti Zaharia. Zhaaria, of course, a famous developer and co -founder of Databricks; He invented its foundational technique while working on his doctorate.

Playerzero announced on Wednesday that it raised a 15 million dollar chain led by Ashu GARG from Foundation Capital, which is early in the early data data. This is followed by a seed of $ 5 million led by Green Bay Ventures and many observed angels, including Zhaaria, CEO of Dropbox Drew Houston, CEO of Figma Dylan Field, and CEO of Vercel Guillermo Rauch.

During his time in Stanford Don, Corata, who is now 26, was working on artificial intelligence model pressure and “actually exposed to language models,” he says. He met the developers who formulated some of the first artificial intelligence coding tools.

“There is this world in which computers will write the code. Humans will not be anymore.” “How will the world look at that stage?”

It was known before the term “AI SLOP” that these agents would have produced a symbol of breaking things as the human supervisors do.

This problem will also exacerbate many agents who come out much more than the code than before. It will not always be that humans from all written instructions from the artificial intelligence of errors or hallucinations. The issue becomes more intense for the large and complex code rules on which institutions depend.

Playerzero trains models “that deeply understand the rules of code, and we understand the way they built, the most wonderful way,” says Koratana.

His technology studies the history of errors, issues and solutions in the institution. When something is broken, its product can be “knowing and fixing the cause, then learning from these errors to prevent them from happening again,” says Korana. Its product is similar to an immune system for large code rules.

Landing Zaharia, his consultant, was a first step to collect donations, but the moment when his idea really proven when he showed an experimental offer for another famous developer: Rauch. Rauch is a founder Triple-Ricorn Developer Tool Vercel and the creator of the famous JavaScript frame Next.js.

Watch Rauch a show of KORATANA with interest but suspicion, asking about a “real” extent. Corata answered that this was a symbol “works in production. Like, this is a real example. It was calm,” says Corata. Then his investor, who will be soon, answered, “If you can already solve this the way you imagine, it is a really big problem.”

Of course, Playerzero is not alone trying to solve the problem of errors created from artificial intelligence. Only last week, anysphere index I released Pogbot To detect coding errors, just one example.

However, Playerzero is already gaining an attractive focus on large code rules. Although it was perceived by a world in which the agents are programmers, it is currently used by many large institutions that use common spectra for coding. For example, Zuora subscription bills is one of Startup Marquee customers. Zuora said technology is used through its engineering teams, including its largest symbol and its billing systems decline.