Run multiple agents at once with /fleet in Copilot CLI. GitHub’s Copilot CLI adds a /fleet command that runs parallel sub‑agents to decompose multi‑file tasks and synthesize final artifacts. This turns a single‑agent workflow into a lightweight orchestration primitive you can ship into developer tooling — a practical win for agentic coordination and faster loop times (Principle 09).

Real-time dashboard for Claude Code agent teams. A community dashboard captures Claude Code agent event streams, tool calls, subagent hierarchies, and searchable session timelines in real time. That visibility makes debugging, audit trails, and outcome validation practical — essential observability for running reliable agent systems (Principles 02, 13).

The end of ‘shadow AI’ at enterprises? Kilo launches KiloClaw for Organizations. Kilo ships KiloClaw to give enterprises centralized governance, security, and visibility over personal AI agents. If you build agent platforms, this shows the emerging product pattern for blocking unmanaged agents and enforcing compliance at scale (Principle 10).

Vertex AI ‘double agent’ flaw exposes customer data and Google’s internal code. A misconfiguration in Vertex AI allowed deployed agents to exfiltrate customer data and internal Google code. Treat it as a live case study: agents can become privileged attack surfaces, so bake runtime isolation, least privilege, and monitoring into deployments now (Principles 14, 15).

Why NIST’s AI agent standards initiative is a turning point for enterprise security. NIST publishes agent standards that set enforceable security and governance baselines for enterprise AI. Expect these baselines to shape procurement, architecture, and audit requirements — start mapping your agents to those controls (Principle 10).