The new Proteus foundation model mimics a security analyst’s workflow, delivering more than 99 % detection of phishing attempts and surpassing conventional rule‑based filters.
AegisAI has introduced Proteus, a groundbreaking foundational model that examines every inbound email with the same rigor a human security analyst would, promising detection rates above 99 % for phishing attempts.
How Proteus Works
Proteus leverages large‑language‑model techniques to parse email content, metadata, and embedded links, then cross‑references these signals against a continuously updated threat intelligence database.
Unlike traditional rule‑based filters that rely on static signatures, the model dynamically adapts to emerging phishing tactics, reducing false positives and catching novel attacks that would otherwise slip through.
Performance Benchmarks
In internal testing, Proteus identified more than 99 % of simulated phishing emails while maintaining a low false‑positive rate, outperforming legacy solutions that typically hover around the mid‑90s percent range.
- Real‑time analysis of every message before delivery
- Context‑aware assessment of URLs, attachments, and sender reputation
- Continuous learning from new threat feeds
- Seamless integration with existing email gateways
Deployment and Compatibility
Proteus is offered as a cloud‑native service that can be paired with on‑premises or SaaS email platforms, providing API endpoints for easy incorporation into corporate security stacks.
AegisAI also announced a partnership program for MSSPs, enabling them to extend Proteus‑powered protection to their client portfolios without additional hardware overhead.
Industry Impact
Analysts view the launch as a shift toward AI‑first email security, where foundational models replace rule sets and signature databases, potentially setting a new standard for enterprise protection against social engineering.
The company plans to roll out additional features, including automated incident response suggestions and integration with security information and event management (SIEM) tools.