Originally published on LinkedIn.
In a recent video, Nate B. Jones offers a sharp articulation of the skills required to navigate the evolution of AI. He calls it "Frontier Operations". Watch the video here.
The core idea: we all know that AI capability is an expanding bubble.
- Inside it — what agents do reliably.
- Outside — what still requires a human.
The real bottleneck isn’t capability — it’s human capacity to operate at that moving boundary.
Nate correctly identifies that the most interesting work happens on the surface of this bubble — and that this surface is expanding.
He names five persistent skills:
- boundary sensing: knowing how big the bubble is - and this changes with every new model !
- seam design: structuring the human↔agent transitions in a clean, verifiable, recoverable manner (this is what we are building Supervaize for !)
- failure model maintenance: maintain a percise understanding of how AI fails today. The key here is precision- general statements are useless...
- capability forecasting: in order to build, we need to visualize the reality that is being shaped
- leverage calibration: where to focus our attention - This is one is tricky !
I would add another one to this list: multi-tasking. When we manage agents, we become the bottleneck of our own processes. As I’m writing this article (co-writing), one agent is drafting new requirements, another is testing and documenting a feature, another is reviewing code, and another is actively writing it. All of them are waiting for me to finish this sentence… <…> I’m back.
Nate also explains why this matters. As organizations try to navigate the AI shift, the ability to operate agents becomes critical. PM agents will help — but by definition, there will always be a frontier. Operating on this frontier is the skill we need to cultivate.
In organizations, this means building practice environments. That's what we built Supervaize for !
Supervaize is the operational studio where business managers experiment, experience, and operate their agents — building familiarity and expertise with those new resources. This isn’t a one-time training. No one becomes a surfer by reading about surfing. Unless you get wet, you’ll never get to ride the waves.
And this is where Supervaize fits. If the frontier is a moving surface, you need a familiar space to operate, you need instrumentation, you need visibility, you need governance and security. You need to know who controls what, what changed, what failed, and why.
The question isn’t whether AI will move the boundary. It will.
The real question is whether your organization has the operational layer to move with it — deliberately, safely, and fast.
Let's finish on another great insight from Nate:
"If your agent hasn’t surprised you recently, you’re not at the frontier."
Follow him if you don't already, most of his videos are simply brilliant: https://www.youtube.com/@NateBJones