Straiker Raises $64M to Secure AI Agents

Silicon Valley startup Straiker has secured $64 million in Series A funding led by Marathon Management Partners to strengthen security for autonomous AI agents. The investment will support GPU infrastructure, AI model pre-training, reinforcement learning, and advanced security research aimed at improving threat detection while maintaining low latency and operational efficiency. The company also plans to expand its global sales, customer support, and marketing operations.

Founded in 2024 by former Palo Alto Networks executive Ankur Shah, Straiker has now raised a total of $85 million. The company is building a security control platform that protects enterprise AI ecosystems, supporting leading coding assistants such as Codex, Cursor, Claude Code, Microsoft Copilot, and ChatGPT Enterprise. Its goal is to provide consistent security across multiple AI platforms, allowing enterprises to adopt agentic AI without vendor lock-in.

Shah believes organizations are rapidly moving from task-based AI assistants to fully autonomous AI agents capable of independently developing applications, deploying infrastructure, executing business workflows, and managing long-running operations. As these agents operate with minimal human intervention, they significantly expand the enterprise attack surface, making AI-native security an urgent requirement.

Straiker is focusing on emerging threats unique to autonomous AI, including malicious "skills files" that can secretly embed harmful instructions within legitimate AI training content. Such attacks can manipulate AI agents into accessing confidential data, leaking sensitive information, or performing unauthorized actions. Detecting these language-based attacks requires specialized AI security models trained on real-world exploit scenarios and continuously refined through reinforcement learning.

The new funding will enable Straiker to move beyond fine-tuning existing open-source models by investing in pre-training and post-training security models. The company plans to replace generic internet datasets with curated security intelligence, enabling its AI to better recognize sophisticated attacks. By continuously learning from simulated cyberattacks and defensive responses, Straiker aims to build adaptive AI security systems capable of protecting the next generation of autonomous enterprise agents.