Pallma

Pallma is a cybersecurity company focused on security for AI agents and autonomous systems. The company develops runtime security controls and boundary enforcement mechanisms for AI agents operating within business-critical environments. Core offerings include: - Runtime control and policy enforcement for autonomous AI agents - Security boundaries to prevent unauthorized actions by AI agents in production systems - An AI Hack Challenge platform (challenges.pallma.ai) where users can practice real-world AI-specific exploits, including prompt injection attacks and data exfiltration techniques against live AI applications The AI Hack Challenge is an interactive, hands-on training environment designed to expose practitioners to offensive techniques targeting AI systems, helping organizations understand the attack surface of deployed AI agents. Pallma's approach centers on the premise that as AI agents gain autonomy in business workflows, they introduce new security risks that traditional security tools do not address. The company targets organizations deploying AI agents in operational or business-critical contexts and aims to provide guardrails that enforce what those agents are permitted to do at runtime. The team cites experience from companies including Microsoft, AWS, Expedia, King, On, and Just Eat. The company is in an early-access stage, offering demo bookings to prospective customers.