Introducing Agent Control: The Open-Source Control Plane for AI Agents

Guardrails exist. They're just in the wrong places. Agent Control is an open-source control plane that lets you govern every AI agent from a single decorator, a centralized policy server, and hot-reloadable rules. No redeployment, no vendor lock-in. In this video, Galileo co-founder and CTO Yash Sheth walks through why current approaches to agent governance are broken, how Agent Control works, and why we're open-sourcing it under Apache 2.0. ⭐ Star the repo: https://github.com/agentcontrol/agent-control 📖 Docs: https://agentcontrol.dev/ 📅 Register for the webinar: https://galileo.ai/blog/announcing-agent-control 📝 Read the launch blog: https://galileo.ai/webinar/agent-control-centralized-governance-for-every-agent Agent Control works with the frameworks and guardrail providers you already use — including LangGraph, CrewAI, AWS Strands, Galileo Luna, NVIDIA NeMo, AWS Bedrock, and Cisco AI Defense. Built in the open with contributions from some of the largest AI infrastructure and security companies in the world. About Galileo Galileo is the agent observability and guardrails company that helps enterprises run trustworthy agents. Learn more at https://galileo.ai Chapters 0:00 — The state of agent safety today 0:18 — Why guardrails are in the wrong places 0:45 — How Agent Control works 1:00 — The @control() decorator 1:09 — Policies decoupled from code 1:30 — One control plane for every framework 1:42 — Any guardrail, no vendor lock-in 1:57 — Why we're open-sourcing it 2:21 — Get involved