5h15hJust shared a few thoughts from #NVIDIAGTC after spending time with the Galileo #AI team.
Agents are starting to act, not just respond.
OpenClaw was a wake up call. Why a control layer is becoming essential: https://www.linkedin.com/posts/shishs_agent-control-the-open-source-control-plane-activity-7441617992448352256-RKdI
Agent Control — The Open-Source Control Plane for AI Agents | ShiSh S.
Walking the floor at #NVIDIAGTC last week, the conversation was no longer just about #chatbots; it’s about #AIAgents that do things. The Galileo (a Microsoft for Startups Pegasus partner) team showed me the "#OpenClaw" post-mortem, and it’s a sobering reminder: when an agent starts reasoning across workflows and hitting "publish" in the real world, "safe prompts" aren't enough. You need a steering wheel. (https://lnkd.in/gnqZGcmb) That’s why their move into 𝐀𝐠𝐞𝐧𝐭 𝐂𝐨𝐧𝐭𝐫𝐨𝐥 (https://agentcontrol.dev/) is so timely. It feels like the early days of Cloud infrastructure all over again: we need a control plane to define policies across tools and workflows. If you’re building in this space, you have to ask: Who is observing the agent? Who is guiding the decisions? Where are the guardrails? 𝑺𝒄𝒂𝒍𝒆 𝒐𝒏𝒍𝒚 𝒉𝒂𝒑𝒑𝒆𝒏𝒔 𝒘𝒉𝒆𝒏 𝒚𝒐𝒖 𝒉𝒂𝒗𝒆 𝒄𝒐𝒏𝒕𝒓𝒐𝒍. Yash Sheth, Heena Purohit, Hans Yang, Tom Davis, Jared Prins, Deepak Jose, Kevin Hague, Matt L. Barker, Ph.D., Faiz Sherman, Jim Andrew, Maria Rigatto
ComputerBase
GioCo
Casper Rutjes
Francisco Villafáñez
Victor Klaue