General Analysis

General Analysis is an AI security company headquartered in San Francisco that provides enterprises with tools to safely deploy agentic AI. The company was founded in 2025 by Rez Havaei, Maximilian Li, and Rex Liu. An earlier report from 2025 also mentions Alan Wu as a co-founder. The founding team consists of researchers with backgrounds from prominent tech companies and institutions like NVIDIA, Cohere, DeepMind, Harvard, and Caltech. Liu and Havaei reportedly met in poker circles, where they bonded over game theory and risk analysis. Their combined expertise in offensive security and AI safety research forms the foundation of the company's approach. The company's mission is to address the security gap created as businesses increasingly adopt autonomous AI agents in high-stakes environments. General Analysis operates on the principle that AI security is an empirical issue that requires rigorous measurement of system behavior under various conditions. In April 2026, the company announced a $10 million seed funding round led by Altos Ventures, with participation from 645 Ventures, Menlo Ventures, Y Combinator, and other strategic investors. This capital is intended to expand operations and further development efforts. The firm's platform offers a comprehensive suite of AI safety tools, including run-time guardrails, red-teaming frameworks, interpretability techniques, and observability. It is designed to systematically identify model failure-modes and improve model robustness. The platform's two main functions are providing a wide range of defensive tools and performing powerful adversarial simulations to benchmark an agent's reliability and security. By training proprietary adversarial agents, General Analysis can map a customer's runtime environment and generate multi-step exploits to uncover system-level risks. This allows them to forecast vulnerabilities and implement state-of-the-art defenses for their clients, which already include enterprise customers in the support and finance sectors. Keywords: AI security, agentic AI, AI safety, model robustness, adversarial evaluation, red-teaming, AI guardrails, model failure modes, enterprise AI, AI observability, AI risk management, autonomous agents, defensive tooling, vulnerability forecasting, runtime controls, empirical measurement, AI agents, enterprise platform, adversarial simulations, AI model safety