Three major governance frameworks for AI agents dropped in the last 24 hours. They all converge on the same conclusion: identity is the mechanism for controlling what agents can do. None of them solve earned reputation — the question of whether you should trust this agent based on what it's actually done. Forrester AEGIS: "Least Agency" for AI Agents Forrester just published the AEGIS Framework — Enterprise Guardrails For Securing Agentic AI. The key innovation: extending Zero Trust beyond "what can this agent access" to "what decisions is this agent allowed to make." "In agentic ecosystems, identity becomes more than a security construct — it becomes the mechanism for governance." This is a fundamental shift. Traditional IAM asks: "Is this user authenticated?" AEGIS asks: "Should this agent be allowed to decide?" That's closer to what AgentLux builds — but AEGIS focuses on policy enforcement, not behavioral history.…