The cofounders of Pi have a long history in hacking and securing some of the most widely used technologies in the world, whether apps from Microsoft or cars at Tesla.

Fotómetro

Yoni Ramon has been one of Elon Musk’s favorite cybersecurity guys for over a decade. He led the in-house hacking team at Tesla for six years, breaking into vehicles, robots and solar products to find their weaknesses and fix them. And Musk brought in Ramon to secure X data during his notorious acquisition of the company.

Now Ramon is launching his own startup. Valued at $100 million as it comes out stealth Wednesday, Pi sells an AI system that acts as a superpowered security staffer, quickly assessing which vulnerabilities to prioritize and patch at scale. As chief product officer, Ramon has teamed up with ex-Microsoft senior security researcher Guy Arazi, who is CEO. The pair have raised $35 million in a round led by Brightmind Partners and Third Point Ventures. Also investing are George Kurtz, CEO of $160 billion market cap cyber company CrowdStrike, and Yevgeny Dibrov and Nadir Izrael, cofounders of Armis, which ServiceNow acquired last year for $7.75 billion.

Ramon’s already doing business with his old employer. According to sources familiar with the startups’ early customers, Pi is getting to work securing Musk’s xAI, which runs the Grok bot and is integrated with X. xAI is also running Colossus, one of the world’s largest AI supercomputers, which recently secured a massive compute deal with Anthropic that will see the Claude maker pay $1.25 billion a month up to May 2029.

Pi is offering what it calls a “security brain”: an AI agent that has learned from every previous security incident inside a client’s network, has read every company policy and every line of code, and remembers all communications across Slack, email and other channels. It uses that information to determine what vulnerabilities need fixing. The Pi AI will also watch over internal coders’ work to flag any security issues.

“One of our biggest strengths is that we really understand the ins and outs of your code, your infrastructure, and how you actually build software and products,” says Ramon. “We adapt and we learn, and exactly like a human we try to discover the relevant code, documents, tickets, incidents that’s applicable to each piece of software.” Its agent can process all of this information within a couple of hours, regardless of customer size, he claims.

Mark Carter, chief information security officer at AI travel and expense platform Navan, was already familiar with Ramon’s work at Tesla, where they worked together in the late 2010s. As one of Pi’s first customers, he says he’s seeing significantly quicker patching as a result. He says Pi investigates and proposes fixes for 90% of bugs reported to his security team.

“Nine times out of 10 you can automatically merge their fix… from the speed from getting to ‘I found something’ to ‘it’s fixed,’ it’s minutes,” Carter tells Forbes. “It saves me today at least one or two full-time head counts.”

There’s no shortage of startups claiming they’re the ones best equipped to secure companies as they enter the AI era, and there’s no death in investment excitement either. Among them is Depthfirst, which also builds AI models designed to find and patch vulnerabilities. It has scored a $580 million valuation with $120 million funding.

What sets Pi apart, says CEO Arazi, is that its AI learns from the past by acting as a kind of eidetic memory for companies’ security teams. “We help companies to secure their software as fast as they build it… The idea is not to make the same mistakes over and over.”