Kai Launches Agentic AI Security Platform With $125M Funding

Artificial Intelligence & Machine Learning
,
Next-Generation Technologies & Secure Development
,
Threat Intelligence

Startup Platform Targets Autonomous Detection and Exposure Management

Michael Novinson (MichaelNovinson) •
March 11, 2026    

See Also: Boost your cybersecurity team capabilities with GenAI

The Evolution Equity Partners funding will help San Jose, California-based Kai create a security platform that combines exposure management, threat intelligence, detection engineering and remediation into a single system capable of operating at machine speed, said co-founder and CEO Galina Antova. The funding will primarily support expansion of the company’s AI research team and engineering workforce.

“We wanted to build as comprehensive and as integrated a platform from day one as possible,” Antova told Information Security Media Group. “That is the ambition. The money really allows us to get access to the best AI scientists, to hire the needed engineers that we have and to make sure that the product is enterprise ready.”

Kai, founded in 2025, has been led since its inception by Antova, who co-founded Claroty in 2015, served as its chief business development officer and remains on the company’s 12-person board of directors. The company’s platform is designed to integrate multiple security workflows into a single system that can operate autonomously, Antova said (see: Another Global Ransomware Outbreak Rapidly Spreads).

How Kai Brings AI Agents to Exposure Management, Detection

Rather than layering new tools on top of existing systems, Antova said Kai is attempting to redesign security workflows from first principles. The platform integrates data sources, threat intelligence, vulnerabilities, asset context and remediation actions into a unified environment in which AI agents can autonomously perform tasks that previously required human intervention, Antova said.

“The security stack is extremely fragmented, but that’s not how the attackers work,” Antova said. “They don’t care what category you’re in or what tools you have. Having more that are poorly integrated actually makes it extremely, extremely challenging to consume.”

Kai’s platform attempts to automate detection intelligence by ingesting threat intelligence reports, extracting relevant tactics and techniques, and correlating that information with the organization’s known vulnerabilities and infrastructure data. Kai uses agentic AI to address areas where machines outperform humans including large-scale data analysis, pattern recognition and rapid decision-making.

“Sifting through terabytes of data that goes into the scene and figuring out the patterns and figuring out what belongs there and what doesn’t, this is not something that the human brain can autonomously do in a matter of seconds, but agentic AI systems can,” Antova said.

Using AI agents, Antova said Kai’s system can analyze large volumes of vulnerability data, eliminate false positives, determine the most effective remediation actions and deploy those fixes automatically when appropriate. By bringing vulnerability analysis, contextual risk assessment and remediation automation into one platform, Kai seeks to fix the bottlenecks that typically slow down vulnerability management.

“That full chain of execution, getting to remediation, instead of something that was months and many team handoffs now becomes a matter of an hour, a couple of hours,” Antova said.

What Humans, AI Agents Are Best Suited for in Workflow

Kai began with several assumptions about how agentic AI could improve cybersecurity workflows, and then engaged with practitioners to validate those assumptions and refine the platform’s capabilities. Because many large enterprises face similar security challenges and infrastructure complexities, Antova said customer feedback requests often translate into features that benefit multiple customers.

“Everything that we built came from multiple customer requirements,” Antova said. “We listened to the pain. We were proposing solutions. We were refining and expanding. A very natural conversation would present what we know will get the adoption.”

Many tasks in security operations such as analyzing large volumes of data or correlating patterns across multiple datasets are well suited to AI systems since machines can perform these tasks much faster, Antova said. Human analysts instead focus on strategic decision-making such as determining defensive priorities, designing security strategies and supervising the actions of AI systems, according to Antova.

“What we absolutely need the humans for is the strategy,” Antova said. “How do you prioritize different defenses? How do you orchestrate it in terms of where you need the agentic AI systems to focus? We wanted all that the machine can figure out to basically be figured out by the machine. That frees up the humans to really think.”

Vulnerability management tools like Qualys, Rapid7 or Tenable may handle scanning, while other tools address threat intelligence, detection engineering or asset management, and human analysts bridge the gaps between these tools through manual processes, spreadsheets and investigations. Kai replaces this fragmented system with an integrated solution that automates many of the underlying workflows.

“How fast can we expand the platform? How fast can we build toward that state of taking over more and more of the security work in an organization?” Antova said. “And then how many of those organizations are we touching?”