These days you can build anything with AI. And it’s genuinely something, what people do with it. Right now I see three types.
The first build text. They have it written, have it polished — or at least polish what they reckon needs polishing.
The second are coders. By now it barely matters whether they’re “real coders” or just “vibe coders” — that distinction is already old news. Code without AI and you’re a dinosaur. Code mostly with AI and you’ve long since lost control of your code (and some have lost control of their lives, too). Anyway. What comes out are more or less complicated programs running on someone’s own machine. A few make it into production, onto a real server. Far fewer. Because the moment your code leaves your laptop for production, everything gets harder, and suddenly nothing runs the way you wanted.
And then the third group, growing by the day. The ones building little dashboards, trying to squeeze their whole complicated working life onto a single screen. One dashboard after another.
And nobody does anything
I see it in my own company. In no time my colleagues caught the magic of colourful dashboards. PowerPoint? Nah, let’s just look at a dashboard together. And now I sit in meetings that get longer and longer, where we stare at complicated dashboards full of tiny numbers, drill into ever more granular data, discuss everything under the sun and derive a hundred action items. Which, in the end, we don’t act on.
The dashboards give us transparency. Far too much of it. And we do nothing.
At first I thought the dashboard builders would be the first to go. Anyone can look at a dashboard, and nobody’s going to pay you to sit and stare at one all day. But it seems I had it wrong. Right now the dashboarders — that’s what I’ll call them — are doing just fine. They’ve quietly turned into the only people who understand their own dashboards. And because the data can always be sliced finer, and that of course needs explaining — and because at that depth nobody really knows why we’re looking at any of it anymore — the dashboarders are having their moment.
Still, I can’t quite see their value. Next to the people who actually work with AI, who produce real output and outcome, the dashboarders contribute fairly little. They’re handy for presentations, and the odd ad-hoc question. But Claude could probably break it all down just as well. Better, even.
The dashboards exist now. The dashboarders shouldn’t get too comfortable; they’d do well to start making things too. Because soon enough the boss will work out that it was never about the dashboard. It’s about the decision hiding in the numbers.
What I also find interesting is what’s happening to the data analysts. For years they were crucial, they built the whole data infrastructure that everyone now taps with one click to build yet another dashboard. And suddenly the analysts look redundant too? I think that’s short-sighted. An analyst looks at the data differently than a marketing person does. You need both views to make the best calls.
Right when it gets interesting
Anyway. I just wanted to get it off my chest: this endless dashboard-gazing gets on my nerves. It leads nowhere. The discussions always stop right when they get interesting. Like yesterday. Another hour clicking through layers of dashboards. Nobody took notes, of course. We must have talked about a hundred action items. What I miss is the conversation about which of them gives the best bang for the buck. We seem buried under all the lovely possibilities, too rattled to decide anything. Because: next week the dashboard will have changed again, or grown another 27 layers, to make things easier for us. Or to keep the dashboarders in a job …
This post first appeared in German on reinergaertner.de, where I’ve been writing since 1997 — back when the internet still had that new-car smell. An AI assistant helped with the translation under my supervision. If something reads a bit odd, blame the Denglish in my head.