The biggest GPU company on the planet just built an enterprise wrapper around a free open-source tool because their customers are already using it — and IT can’t stop them.
At GTC this week, NVIDIA announced NemoClaw — a software stack that deploys secure AI agents on OpenClaw with a single command. Enterprise guardrails, audit trails, sandboxed execution. It’s a good product, and it’s a confession.
I talk to organisations every day that are trying to build enterprise agents. To be clear: the first wave of enterprise AI worked. Chatbots, RAG pipelines, hybrid search — organisations have genuinely nailed that stuff, and it’s delivering real value. But agentic is a different problem entirely. Outside of coding — which works, and works well — most organisations are struggling to build agents that actually do things autonomously on behalf of users in a way that changes how the business operates. The demos are fine. The production value is… not. OpenAI just declared a “code red” and pivoted its entire company toward coding and enterprise because that’s the one agentic use case that’s actually delivering. That should tell you something about where we are.
Meanwhile, individuals building personal agents are quietly getting results. I’ve been thinking about why, and I keep coming back to something that feels a lot like shadow IT — except it’s not quite that.
Shadow IT was people using unapproved tools. Dropbox instead of the network drive, WhatsApp instead of the approved messenger. You were still a consumer. Someone else built the thing. What’s happening now is different — people are building their own tools, without developers, without a ticket system, without waiting for roadmap prioritisation.
That’s not shadow IT. That’s shadow engineering.
It also explains why the personal agent is arriving well ahead of the enterprise one. Enterprise agents are stuck — not just because of procurement friction, though that’s real. The deeper problem is that the person building the agent isn’t the person using it. You end up with something built around what someone thinks users want, filtered through a product spec, a security review, and whatever made it onto the roadmap this quarter. The feedback loop is slow, the intent is approximate. With a personal agent, you’re the builder and the user. You notice it’s doing something annoying and you fix it that afternoon. The intent is perfect because it’s yours.
I’m already doing this. I use my own agent for account research before customer calls, to track interactions with key contacts, to surface things I’d otherwise miss. I have built my agent to be shaped around my actual job, doing the things generic enterprise AI solutions don’t quite get to. I’m probably not alone in that.
Here’s what I think happens next: BYOA. Bring your own agent. And look at what NVIDIA just announced — rather than IT building the agents people need, NemoClaw gives individuals enterprise-approved infrastructure to build their own.
The organisation sets the guardrails. The individual defines the use case.
That’s a fundamentally different model to how enterprise software has ever worked, and it’s the smart read on where this is going. Not the enterprise dictating the AI use case to employees. Employees deciding what they actually need, inside a sandbox the enterprise can live with.
IMO the winners will be the organisations that give people the tools to build what they need, not the ones who try to build it for them.