Ask a security analyst what went wrong with the first wave of AI tooling in the SOC and the answer rarely involves accuracy benchmarks. It involves the sinking feeling of reading a machine-generated verdict and having no way to check it.
That specific problem, more than raw alert volume, is what Mate Security says it built its platform to fix. As reported by Axios, the company has now raised a $35M Series A led by Canaan Partners, with participation from Insight Partners, Team8, and M12, Microsoft’s Venture Fund. Total funding stands at over $50M.
The Trust Deficit
Founded by Wiz and Microsoft veterans, Mate addresses a critical cybersecurity challenge: security operations architecture is not designed for the speed and dynamic nature of AI-scale attacks. The consequence for practitioners is compounding rather than linear.
Many of today’s AI security solutions have failed to earn the trust of security teams, leaving analysts overwhelmed not just by a higher volume of alerts, but with AI outputs they cannot verify or act on with confidence. Anyone who has run a Tier 1 queue understands the arithmetic. A tool that produces a confident conclusion without a traceable path to that conclusion adds work. The analyst still has to reconstruct the reasoning before signing off.
Mate was built to solve that.
Context as the Missing Input
Mate provides an open, agentic security operations platform that enables organizations to contain AI-scale attacks. Underneath sits a patent-pending context layer designed for agent precision, powering specialized agents that detect, investigate, respond to, and hunt for threats based on a deep understanding of the customer’s business.
The premise is that most bad verdicts are not model failures. They are knowledge failures. Mate builds an organizational “brain” equivalent to the collective knowledge of an experienced, elite cyber defense team, with a deep understanding of how the organization operates. As a result, Mate can make precise and fast verdicts.
Consider the two cases the company uses to demonstrate this. When a security alert is raised for multiple suspicious login attempts, Mate will know whether security testing was planned during this time, and will report this as a likely non-threat. If an employee downloads multiple sensitive files, Mate understands the broader organizational context, including personnel changes and document classifications, to accurately assess whether the behavior is a genuine threat.
Every experienced analyst already performs those checks. They check the change calendar. They check whether the employee resigned last week. They check the sensitivity label on the files. The senior analyst’s advantage has always been institutional memory, not superior pattern matching.
Permissions, Memory, and Restraint
Investing in security know-how alone is not sufficient to build trustworthy AI security at scale. Mate invests heavily in bringing frontier-class AI for security. Mate’s agents run on persistent memory that compounds with every investigation, communicate through structured agent-to-agent protocols, and operate under least-agency principles, meaning that each agent only gets the permissions and context that its task requires.
Least agency will read as familiar to anyone who has fought through an identity governance project. It is least privilege, applied to autonomous software rather than human accounts. The persistent memory piece is the one that changes the shape of the product over time. An agent that remembers the outcome of last quarter’s investigation is a different asset than one that starts cold on every ticket.
“When we started Mate, we knew we had to invest in the foundation: context and trust, and build them deeply into our product,” said Asaf Wiener, CEO and Co-Founder of Mate Security. “We brought in some of the best AI builders and security experts, and I’m excited to see how well this approach is being received by the market. We will continue moving fast and stay laser-focused on our customers, as we expand into new markets and categories to build the Open Security Operations foundation of the future.”
Governed Context, Not a Data Lake
Mate provides a single, open place where it collects, resolves, cleans, and maintains the customer’s business knowledge in a Security Context Graph, then lets AI security operations run on that governed context managed by Mate’s orchestrator agent. This foundation allows Mate agents, best-of-breed vendor subagents, and customer-built agents to encode deep organizational expertise, while Mate’s trust mechanisms enforce permissions, quality, coherence, auditability, and earned autonomous remediation and response.
There is one design decision buried in that description worth surfacing for practitioners. Mate queries evidence across the customer technology stack, at the source, and lets each agent use the same governed context and controls, so customers can extend AI security operations without fragmenting trust, reasoning, or response. Querying at the source is a different posture than centralizing everything first. It sidesteps a familiar failure pattern where the copy drifts from the original.
The Adoption Signal
The funding comes as an increasing number of Fortune 500 enterprises adopt Mate as their agentic security operations platform. Mate has grown by over 500% since Q3 2025, with the new capital intended to help the company meet accelerated demand.
Large enterprise SOCs are not early adopters by temperament. A 500% growth figure inside that buyer profile suggests the evaluation criteria are being met somewhere other than the demo.
Canaan’s read supports that. “AI is forcing a fundamental rethink of security operations. What stood out to us about Mate wasn’t simply its use of AI; it was the team’s conviction that trustworthy AI requires a deep understanding of how an organization operates,” said Joydeep Bhattacharyya, General Partner at Canaan. “By building a shared context layer that gives AI agents that understanding, Mate has taken a fundamentally different approach to security operations. The customer feedback and success we’ve seen in competitive evaluations reinforce our belief that the team is solving an important problem in a differentiated way.”
Who Else Is Backing It
“Security operations was not built for the speed or scale of modern AI-driven attacks,” said Teddie Wardi, Managing Director at Insight Partners. “Mate is doing something few security companies have managed: combining genuine AI depth with operational trust to rebuild security operations for the AI era. We are proud to support a team that consistently outexecutes.”
“The pace in cybersecurity right now is faster than anything we’ve seen,” said Ori Barzilay, Partner at Team8 Capital. “As one of the leading cybersecurity venture funds, we have a clear view of what exceptional looks like, and Mate still shines above the rest. Their business traction, product development, and talent hiring are exceeding every benchmark for a company at their stage, even in the agentic era.”
“Mate is bringing frontier-level AI models to cybersecurity, and its exceptional growth reflects a defining shift we see in the market,” said Todd Graham, Managing Partner at M12, Microsoft’s Venture Fund. “Organizations want the freedom to adopt the latest AI advances without being locked into a single model or vendor. Mate’s open platform delivers exactly that for security operations, and we’re excited to partner with the team on this journey.”
Who Built It
The team behind Mate combines experienced AI builders, who shipped production LLM systems, with cyber defenders who ran investigations in the world’s largest organizations. Mate’s AI experts specialize in high-stakes business and security decisions. They’ve held senior AI product and research leadership positions at companies including Meta and Microsoft, and include alumni of Cornell University, the Weizmann Institute, the Israeli Technion, IDF Unit 8200, and more.
Mate will be at Black Hat USA between August 3 and 6, 2026, in Booth 4717.
For teams evaluating agentic tooling this year, the questions worth bringing to that booth are narrow ones. What exactly does the agent know about my environment. Who granted it that knowledge. Can I see the trail afterward.
