CalypsoAI

CalypsoAI provides an inference-layer security platform that governs and protects enterprise use of large language models and AI agents at runtime. Its Inference Platform sits between users, applications, and models, enforcing customizable guardrails that block prompt injection, jailbreaks, data-loss and PII exposure, malicious code, and policy violations across any model or provider, while giving administrators centralized visibility, auditing, and access controls over AI usage. The company markets an offensive testing capability it calls Agentic Warfare for continuously red-teaming deployed systems, and reports detection accuracy above 96% with sub-100-millisecond latency suited to production traffic. Founded in 2018 and headquartered in San Francisco with offices in Dublin and Seattle, CalypsoAI grew out of work securing AI for defense and government customers and raised over $40M from investors including Paladin Capital Group, Lockheed Martin Ventures, and Hakluyt Capital, including a $23M round led by Paladin in 2023. In September 2025, F5 agreed to acquire CalypsoAI to fold its inference-layer AI security into F5's application-security portfolio. CalypsoAI is aimed at security and risk leaders at large, often regulated, enterprises that need model-agnostic runtime controls plus governance and continuous adversarial testing across a diverse AI estate.