AegisAI, an email security company founded by members of Google's core security group, raised $36 million in a Series A led by Battery Ventures, with existing investors Accel and Foundation Capital also participating. The San Francisco company builds its own large language models to deploy autonomous agents inside the inbox that interrogate message intent rather than scan for known-bad signatures.
The round brings total funding to $49 million, less than a year after the company emerged from stealth. Cy Khormaee and Ryan Luo, who built reCAPTCHA, Safe Browsing and Web Risk at Google before founding AegisAI in 2025, will use the capital to scale the agent fleet, expand enterprise sales and accelerate the general availability of Vanguard, a product announced in March that extends threat detection beyond the inbox.
For decades, targeted spear phishing at scale was a capability limited to nation-state actors. Crafting a credible lure required researching the victim, mapping their professional relationships and impersonating a trusted contact. That work demanded a skilled operator and significant time. Off-the-shelf AI tools now perform the same reconnaissance and lure-crafting at what AegisAI describes as the cost of a cup of coffee, at unlimited scale.
AegisAI's State of the AI Threat in Email study, presented at the 2026 M3AAWG conference and based on analysis of more than 20,000 phishing, scam and malware emails, found that AI-generated spear phishing grew from 2.8% to 13.9% of all observed phishing attacks in 2025. Those messages reached the inbox more than half the time, evading traditional filters at nearly double the rate of human-written attacks. Of successful AI attacks, 72.6% passed standard email authentication, sent from compromised legitimate accounts with established sending histories.
The FBI's 2025 Internet Crime Report puts a dollar figure on the damage. Total reported cybercrime losses reached a record $20.8 billion. Phishing complaint volume stayed roughly flat, but the losses those complaints represent surged more than 200%, from roughly $70 million to more than $215 million in a single year. Business email compromise, which exploits identity rather than software vulnerabilities, produced $11.64 billion in losses. Ransomware and malware combined accounted for less than $52 million: a 365-to-1 ratio.
"The most immediate, catastrophic risk to your organization isn't an AI agent hacking your firewall. It's an AI model manipulating someone in your organization into handing over the keys, often through the most trusted, most vulnerable contact of the person it's targeting," said Khormaee, co-founder and CEO of AegisAI. "You cannot patch human trust. If your security program still relies on template-based phishing tests and awareness training, you are training your people to spot last year's threat, not a capable agent crafting a novel lure just for them. When the attack is AI, the defense has to be AI."
AegisAI's agents evaluate the intent and identity behind each message rather than matching patterns from past attacks, catching emails that pass every technical filter. The company reports the approach cuts false positives by up to 90% compared to traditional solutions. When a message is flagged, Vanguard follows suspicious links and attachments across the open web as a user would, navigating cloaked pages and weaponized documents, and returns a complete threat report in minutes.
Since its public launch in September 2025, AegisAI has deployed across dozens of fintech and technology companies. At Mesh, a crypto payments company, the agents catch AI-generated spear phishing and business email compromise without the security team managing rules. LangChain, a company prominent in the AI industry, uses AegisAI given the exposure created by its public-facing engineers and executives. At Lokker, the platform caught an attack that came through compromised Salesforce infrastructure, exploiting a trusted vendor's own systems rather than a malicious link or attachment.
"Email is where enterprise trust lives and generative AI just broke every assumption legacy email security was built on," said Dharmesh Thakker, general partner at Battery Ventures, which led the round. "When attacks are machine-generated, personalized and indistinguishable from legitimate mail, the only viable defense is an equally capable AI operating at machine speed. Cy and Ryan spent a decade building exactly these systems at Google scale. They're the team we'd back to win this market."
Vanguard, announced in March, is available to select customers as AegisAI works toward general availability.
