Koi Raises $48M to Safeguard AI Models, Code and Extensions

An endpoint security startup led by a former Israeli Military Intelligence soldier raised $48 million to secure extensions, AI models, code packages and containers.

See Also: AI Agents Demand Scalable Identity Security Frameworks

Washington D.C.-based Koi said the seed and Series A funding will be used to expand support across more non-binary platforms and service large global enterprises, according to co-founder and CEO Amit Assaraf. He said Koi’s differentiation lies in its ability to map, assess risk and govern the complex, distributed software landscape at enterprise scale, making it an early mover in a rapidly growing market.

“Non-binary is something that’s new in the past five years, and now, it’s probably exponentially more dominating than traditional binaries,” Assaraf told Information Security Media Group. “Part of R&D and product involves supporting more and more of these non-binary types of software, like extensions, AI models, MCPs, containers, code and OS packages. Each one of them has its own needs.”

Koi, founded in 2024, employs 40 people and tapped Battery Ventures, Team8, Picture Capital and NFX to lead its latest funding round. The company has been led since inception by Assaraf, who spent four years as an Israeli Military Intelligence soldier focused on cybersecurity and more than five years as chief technology officer of real estate investing app Lando before establishing Koi.

Why Non-Binary Software Is Such a Big Risk

Instead of beginning with a product pitch to CISOs, Assaraf said the technical founders initiated a white-hat hacking experiment, which resulted in the successful compromise of over 400 companies, including major players such as Oracle and Pizza Hut. The strong reaction from CISOs at those companies prompted the team to build an early risk engine prototype, which fueled demand for sophisticated features (see: Cryptohack Roundup: Do Kwon Pleads Guilty in $40B Fraud Case).

“We had a lot of opportunities to do it before, but everything aligned to do it now,” Assaraf said. “And there’s a big opportunity here because we’re probably the only product in the market that can do what we do. We’re the only product that can integrate into an enterprise and map out all of these non-binary software items, give risk and then help prevent. So there’s a big opportunity to build something big.”

Non-binary software might run within the browser, reside in isolated development environments, or execute dynamically from other programs, which Assaraf said leaves organizations blind to its presence and risk. Koi developed a risk engine powered by large language models capable of reading, analyzing and classifying these software types at scale, which Assaraf said will help enable centralized governance.

“Non-binary software is something that’s traditionally been a blind spot,” Assaraf said. “When you look at all the endpoint security tooling and MTMs and other tools that were meant to help you manage the endpoint, they were built in an era where binary software was all there was.”

Koi’s road map includes expanding platform coverage incrementally based on customer need and usage patterns, adding support for tools such as WinGet, a Microsoft package manager, Assaraf said. Each type of non-binary software – be it browser extensions, machine learning models, OS packages or containers – has distinct technical characteristics, requiring tailored solutions for detection and risk scoring, he said.

“We’re significantly scaling it, whether its sales leaders and reps, SDRs, BDRs, the whole creating go-to-market machine that can help us grow really, really fast,” Assaraf said. “We really believe we can build something really, really big to the likes of Wiz.”

What Sets Koi Apart From Competitors

Assaraf said Koi managed to achieve substantial growth with minimal outbound effort, as 70% of its sales pipeline was driven by inbound interest from major companies. The company wants to build a repeatable and scalable sales machine capable of handling both mid-market and large enterprise opportunities across verticals and geographies, according to Assaraf.

“70% of our pipeline is inbound, but we can’t even handle that alone,” Assaraf said. “So we’re taking that money to build a very strong and large go-to-market team.”

While many cybersecurity startups focus narrowly on either SMB or enterprise, Assaraf said Koi serves everything from 100-employee Web3 companies to multinational banks with 150,000 endpoints. Many customers view Koi as a tier-1 vendor, especially in environments like cryptocurrency and Web3 where extensions and packages can directly compromise financial systems or intellectual property, he said.

“We also see Web3-oriented companies,” Assaraf said. “They find Koi is one of the most important vendors they have. And they often pay more for Koi than EDRs.”

Assaraf said some early-stage companies are starting to build products that tackle individual non-binary use cases such as visibility into a single package manager, governance over browser extensions, or gaining visibility into Homebrew installations. But no rival offers comprehensive risk assessment, governance and visibility across all non-binary software types, and all endpoints, users and geographies, Assaraf said.

“There is no vendor that says, ‘Non-binary software as a whole,’ is our focus,” Assaraf said. “So, we’re the first movers on that front. But I don’t think it’s going to be for a very long time. We see startups that are now building themselves to solve kind of the same problem and different cybersecurity vendors that might solve some use cases.”