Exclusive: Mate Security launches Gamebooks to govern how AI agents run investigations

Mate Security, a security operations startup, has introduced Gamebooks, a structured set of investigation procedures for its AI agents. The company aims to address two issues in security automation: broken playbooks when environments change and overly cautious human oversight of critical actions. Gamebooks describe investigative intent rather than execution paths, allowing agents to determine how to proceed based on evidence and conditions.

This separation of intent from execution enables agents to adapt without needing constant updates when tools change. Gamebooks are written in natural language and can incorporate proprietary tools and data. Analyst decisions are preserved in a Security Context Graph, facilitating continuity even with staff turnover. Customers can create their own Gamebooks, or adapt existing playbooks.

Mate claims the architecture ensures AI can reason and adapt while remaining anchored to an organization's specific investigative practices. The company demonstrated the importance of this approach with the July Hugging Face intrusion, where AI-driven models bypassed security measures. Gamebooks is now generally available on the Mate platform and will be showcased at the upcoming CrowdStrike Fal.Con conference. Mate was founded in 2025 by security industry veterans and has raised over $50 million in funding.