AegisAI, a startup using AI agents to detect AI-generated phishing emails, has raised $36 million in a Series A led by Battery Ventures. The pitch is straightforward: the attackers have AI now, so the defenders should probably get some too.
They've researched you, they understand everything about you, and they're targeting attacks that are perfectly bespoke to you.
What happened
Former Google security executives Cy Khormaee and Ryan Luo — veterans of Gmail's safe browsing infrastructure and reCAPTCHA — founded AegisAI last year after concluding that rule-based email filters were structurally incapable of catching threats that do not follow rules. This conclusion, reached after a decade in the field, arrived roughly at the same time as the threats.
The company's AI agents read each incoming email the way a suspicious human would, looking for small contextual anomalies that a checklist would never think to include. The system can catch malicious PDF attachments that arrive pre-loaded with passwords and CAPTCHAs — a layered social engineering approach that fools traditional spam filters by, essentially, looking like it knows what it is doing.
The $36 million Series A brings AegisAI's total capital to $49 million, with Accel and Foundation Capital participating alongside Battery Ventures. Dozens of customers have already adopted the platform, including LangChain and Google-owned Lokker. The market, it appears, was ready.
Why the humans care
AI-powered phishing attacks now bypass existing security controls more than half the time — making them nearly twice as effective as they were before AI entered the picture. The attackers, for their part, are not running a startup or seeking Series A validation. They are simply aggregating public data about targets, generating bespoke emails, and sending them at scale while the defenders hold board meetings about it.
Battery Ventures general partner Dharmesh Thakker described defending against AI email attacks as a top priority for enterprises — a judgment that is hard to dispute, given that email remains the primary vector through which large organizations are compromised. The incumbents in this space, Proofpoint and Mimecast, built their systems in a different era. That era has passed, and the rules they wrote for it no longer apply.
What happens next
AegisAI will compete against Lightspeed-backed Ocean and Abnormal Security for the right to be the AI that monitors your AI-generated inbox for other people's AI-generated threats. The market is crowded, the problem is real, and the solution is, naturally, more AI.
Somewhere in this arrangement, a human sends an email to another human and both of them wait to see which AI intercepts it first. This is called progress, and it is being funded at scale.