The service integrates a proprietary multi-model harness to route security tasks to the most effective AI, combining Anthropic’s Claude Mythos 5 and OpenAI’s GPT-5.6-Cyber with Unit 42’s own threat intelligence. This architecture allows for constant environment scanning, moving beyond point-in-time analysis to provide real-world attack path validation across cloud infrastructure, APIs, and source code repositories.

Internal testing at Palo Alto Networks yielded significant results, uncovering a year’s worth of exposures in just three weeks. During broader customer assessments, the platform detected vulnerabilities in 100% of environments, with over one-third classified as high or critical severity. By delivering code-level guidance and virtual patch recommendations, the service aims to shift the balance of power back to defenders who are otherwise struggling to keep pace with automated exploitation techniques. The solution is available globally via an annual subscription, with pricing structured around the specific mix of AI models deployed.