Google Security Veterans Launch AegisAI with $13M Seed Round to Reinvent Email Security

More than 90% of successful cyberattacks begin with phishing emails, making the inbox the single most exploited entry point into organizations today. This statistic alone highlights the urgency for businesses to rethink how they protect their most common communication channel. Into this gap steps AegisAI, a startup founded by Google alumni Cy Khormaee and Ryan Luo, which has just launched publicly with a $13 million seed round led by Accel and Foundation Capital.

Khormaee and Luo spent nearly a decade at Google safeguarding billions of users through Safe Browsing and reCAPTCHA. With AegisAI, they are bringing that scale of defense to the enterprise, but with a very different philosophy: instead of teaching employees to spot scams or maintaining rule-based filters, AegisAI relies on autonomous AI agents that neutralize attacks before they ever reach a user’s inbox.

Why Traditional Defenses Are No Longer Enough

For decades, email security has leaned on static rules, signature databases, and reputation scoring to identify malicious senders or suspicious content. While these approaches worked in an era of predictable, mass-produced spam, they struggle in today’s world of adaptive, AI-powered attacks.

Modern adversaries now generate phishing lures with large language models, producing messages indistinguishable from authentic communication. These messages are not only linguistically sophisticated but are often personalized, dynamically adapting tone, phrasing, or embedded graphics to each target. Research has shown that AI-generated phishing messages have click-through rates several times higher than their human-written counterparts.

Equally concerning is the trend of attackers abusing trusted services like Salesforce, Zoom, or Google Docs. By embedding malicious content inside platforms that employees use daily, adversaries bypass filters that flag unknown or untrusted domains. The combination of personalization, contextual awareness, and trusted delivery channels has left static defenses a step behind.

A Machine Learning-First Approach to Email Defense

AegisAI addresses these gaps with a fundamentally different architecture: a network of autonomous AI agents that continuously inspect, analyze, and act on incoming email traffic. Each agent is specialized in monitoring a particular signal — from link behavior and attachment content to QR code anomalies and metadata consistency. Together, these agents form an orchestrated defense system that evaluates emails holistically, in real time.

At the heart of AegisAI’s platform is a machine learning pipeline trained on both real-world attack data and adversarial simulations. The models are designed to detect subtle patterns that humans or rules would miss: the statistical quirks of auto-generated language, the minute timing differences in spoofed domain traffic, or the unusual sequence of redirects that signal a phishing kit. By applying ensemble learning, the system combines multiple models’ predictions to minimize blind spots.

Crucially, these models are not static. AegisAI employs continuous learning loops that retrain on new adversarial behaviors observed across its customer base. This enables the platform to evolve as quickly as attackers, who often tweak their campaigns by rotating images, rewriting content, or adding fabricated supporting materials. The company reports that this architecture has already cut false positives — legitimate emails flagged as malicious — by up to 90% compared with legacy solutions.

Autonomous Response and Operational Simplicity

Another defining characteristic of AegisAI’s system is its autonomy. Traditional security tools often flood IT teams with alerts, leaving analysts to decide which messages are safe, which are malicious, and what action to take. This slows response times and introduces human error.

AegisAI instead automates the full lifecycle of threat detection and mitigation. When its agents flag an email as malicious, the system can quarantine or neutralize the threat instantly, without requiring manual intervention. If the threat is ambiguous, the system escalates it intelligently, presenting the SOC team with distilled intelligence rather than raw alerts.

Deployment is equally streamlined. By integrating directly with Microsoft 365 and Google Workspace via API, AegisAI avoids the complex configurations and policy tuning that gateways traditionally demand. Customers are able to activate protection in hours, not weeks, while maintaining compliance through enterprise-grade encryption and data minimization.

Voices of Validation: Customers and Investors

Early adopters have already seen the difference. Mesh CEO Bam Azizi described how AegisAI “stopped attackers in their tracks” and identified everything from fuzzing attempts to AI-generated spear phishing, all without his team managing complex rules. Lokker’s Ian Cohen noted that the platform gave immediate visibility into targeted threats against accounting, engineering, and executive teams, allowing attacks to be contained before damage could occur.

For investors, the appeal lies both in the technology and the founding team’s pedigree. Eric Wolford, Partner at Accel, emphasized that Khormaee and Luo are “AI-native,” with a career-long track record of building scalable AI defenses rather than simply adopting the latest trend. Foundation Capital echoed the view that AegisAI represents the future of enterprise-grade cybersecurity, where adaptive intelligence replaces brittle rule sets.

A Market Ripe for Reinvention

The timing could not be more significant. The email security market is already worth billions and is expected to nearly double in the next several years as organizations shift to cloud-first infrastructure and face mounting pressure to defend against AI-enabled attackers. Business Email Compromise alone costs enterprises billions annually, making it one of the most financially damaging forms of cybercrime.

In this context, AegisAI’s machine learning-first approach positions it not just as another security vendor, but as a potential category-defining player. By combining scalable AI expertise with seamless integration into existing enterprise platforms, the company is aiming to make email defense as adaptive and invisible as the threats it counters.

Looking Ahead: Toward a New Era of Email Security

The road ahead for AegisAI involves expanding its engineering team, accelerating product development, and scaling go-to-market operations. But beyond corporate growth, the company is setting out to redefine what email security looks like in the age of AI.

As attackers weaponize machine learning to create polymorphic phishing kits, deepfake impersonations, and AI-driven social engineering campaigns, static defenses will only fall further behind. Autonomous agents that learn, adapt, and coordinate in real time represent a future where defenses move as quickly as offenses.

AegisAI envisions email not as the weakest link in enterprise security, but as a fortified line of defense — one that continuously strengthens itself through learning and intelligence sharing. If successful, the company could mark the beginning of a new era where AI is no longer a threat multiplier for attackers alone, but a decisive advantage for defenders as well.