Design leaders spend a lot of time telling teams to experiment with AI. Nathan Lavertue, a Design Program Director for IBM Z and LinuxONE, turns that advice back on leadership itself:

We spend a lot of time helping designers understand how to work with AI. The question I keep coming back to is simpler. How many of us are doing the same for ourselves in ways that meaningfully support the business?

So instead of just encouraging my teams to experiment with AI tools, I put myself in the work. I built a design program signals website using IBM Bob. What started as a wireframe to sketch out an idea became something I realized I could actually build myself. That surprise is the whole story.

I appreciate the reciprocity here. If designers are being asked to work through this shift in public, leadership cannot treat AI as a strategy deck it reviews from a comfortable distance. You cannot build useful judgment about these tools by asking other people to absorb the uncertainty for you. That is why I’ve been reading about them, writing about them, and experimenting with them on my own. Whether it’s OpenClaw, Hermes, running a local LLM, ComfyUI, or Claude Design, curiosity is key here.

The interesting part of Lavertue’s example is not that he made a dashboard. Dashboards are cheap. The useful part is that he used AI to make a leadership problem legible enough to discuss. His signals site pulled together team health and business impact, then sorted indicators into required, expected, and optional categories so the absence of a signal became something to interpret, not just a blank cell to punish.

Lavertue is clear about this:

I had to remind myself of that more than once while building it. The signals site was useful. Bob was a capable collaborator. But the risk with any tool that comes together quickly is mistaking the build for the point. The site was never the outcome. It was infrastructure for conversations. Design’s impact on the business was the outcome. Keeping that distinction clear required the same discipline I would ask of any designer getting excited about a new tool.

The site did not replace leadership judgment. It grounded it. Instead of reacting to delayed updates or anecdotal signals, I could engage teams with shared context and a clearer ability to look forward rather than back. This was another form of walking the walk. Not just encouraging teams to work differently but building the system that made that work visible and meaningful.

That feels like the better bar for AI-native leadership. Not “leaders should code now.” Not “every management problem needs a custom tool.” The bar is whether leaders are willing to put their own work through the same change they are asking from their teams.