Armadin has raised a $255.5 million Series B at a valuation above $2.5 billion for a security platform that tries to break into enterprise systems before real attackers do. Instead of sending a small human team through a periodic penetration test, the startup gives enterprise security teams an always-on swarm of AI agents searching for weaknesses in parallel.
The workflow begins by mapping an organization’s internal systems and users, then scanning those assets inside a sandbox designed to contain potentially harmful activity. Armadin says its agents can combine several weaknesses into an attack path—the sequence an intruder could follow to reach sensitive systems—before giving the customer remediation guidance and prioritizing the most severe findings.
That distinction matters because an isolated software flaw may not reveal how much damage an attacker can actually do. In one customer assessment described by Armadin, 26,000 agents scanned more than 25,000 assets and identified 38 verified attack paths. Those are company-reported results, but they illustrate the product’s intended scale: test many routes simultaneously, then give a security team a shorter list of chains to break.
Andreessen Horowitz and Accel co-led the round. Bain Capital Ventures, Redpoint, 8VC, Ballistic Ventures, GV, In-Q-Tel, Kleiner Perkins and Menlo Ventures also participated. The financing follows a $190 million Series A completed about six months earlier and brings Armadin’s total funding to roughly $445 million.
Founder Kevin Mandia previously built breach-investigation company Mandiant, which Google acquired for $5.4 billion in 2022. Armadin applies that offensive-security perspective continuously: the product simulates coordinated attackers, while its enterprise and government customers use the resulting attack paths to decide what to repair first.
Armadin says it is already running agent-led attack campaigns in production for Fortune 500 companies and government customers. The company plans to use the new capital to expand the platform, research, AI training and go-to-market operations.
Published versions remain available when an article is updated or corrected.
- Revision 1 · Initial publication
Initial publication