Arcjet Launches Runtime Security for AI Agents

Arcjet Labs has launched agent runtime security, giving teams real-time visibility and policy controls over AI agents running in production. The system tracks every action, logs all runs, and can halt or escalate calls before they execute, without requiring code changes. The platform integrates with OpenTelemetry, Claude, OpenAI, LangChain and others, and supports on-premise data storage for sensitive environments. Policies run on Open Policy Agent and can be updated without redeployment. Arcjet, backed by Andreessen Horowitz and Plural, now has over 500 production applications using its software.