Armadin Raises $255.5M Series B to Replace Pen Testers with Agent Swarms
Six months, two rounds, $445 million. Armadin, the new company from Mandiant founder Kevin Mandia, announced a $255.5 million Series B on Thursday at a valuation above $2.5 billion. Andreessen Horowitz and Accel led. Bain Capital Ventures, Redpoint, 8VC, Ballistic Ventures, Google Ventures, In-Q-Tel, Kleiner Perkins and Menlo Ventures joined. The $190 million Series A closed only in March.
The product kills the annual pen test. In the old model, a company hires a team of human attackers once or twice a year and gets a PDF of findings. Armadin runs swarms of attacking agents that never stop. They chain small vulnerabilities into a working break-in, the way a real attacker would, so the company can patch before someone else finds the same path. TechCrunch's phrasing of the threat model is telling: bad guys, "or even AI labs with rogue agents."
The timing explains the price. A day later, TechCrunch reported that Epic, which makes MyChart and holds the software layer over 320 million US patient records, has paused most product development after Anthropic's Mythos turned up flaws that could expose patient data. Epic's security chief told the New York Times some configurations could let outsiders read records without leaving a trace in the logs. When one frontier model can make a company the size of Epic stop shipping, continuous automated offense stops being a nice-to-have.
Mandia sold Mandiant to Google for $5.4 billion in 2022, so investors are paying for the operator as much as the idea. The open question is the same one every agent-offense company faces: defenders buy findings, but agent swarms produce a lot of them. Whoever ranks the 3 real kill chains above the 300 theoretical ones wins this market.
Link: techcrunch.com (Armadin Series B, Oct 1)
← Back to all articles
The product kills the annual pen test. In the old model, a company hires a team of human attackers once or twice a year and gets a PDF of findings. Armadin runs swarms of attacking agents that never stop. They chain small vulnerabilities into a working break-in, the way a real attacker would, so the company can patch before someone else finds the same path. TechCrunch's phrasing of the threat model is telling: bad guys, "or even AI labs with rogue agents."
The timing explains the price. A day later, TechCrunch reported that Epic, which makes MyChart and holds the software layer over 320 million US patient records, has paused most product development after Anthropic's Mythos turned up flaws that could expose patient data. Epic's security chief told the New York Times some configurations could let outsiders read records without leaving a trace in the logs. When one frontier model can make a company the size of Epic stop shipping, continuous automated offense stops being a nice-to-have.
Mandia sold Mandiant to Google for $5.4 billion in 2022, so investors are paying for the operator as much as the idea. The open question is the same one every agent-offense company faces: defenders buy findings, but agent swarms produce a lot of them. Whoever ranks the 3 real kill chains above the 300 theoretical ones wins this market.
Link: techcrunch.com (Armadin Series B, Oct 1)
Comments