From months to minutes

The first time I built a marketing data warehouse I had absolutely no idea what I was doing.

I found a tool on the shelf called Funnel that was abundantly paid for and barely used, and figured out that an out-of-the-box API integration is basically a two-step process. That was enough to get started. I was managing a performance team in a multi-billion company with zero reporting, so I became pathologically obsessed. Harmonising dimensions, cleaning anomalies, pulling in more sources, stitching everything together. Then I got to visualisation and that was it. To this day, very few things outside leadership get me more excited than good dashboards.

But now things are getting beyond surreal. If you connect Anthropic's Claude to your data sources, what used to take months now takes minutes. It looks better and, if we're honest, it's probably more accurate than what I would have built back then. There's still some craft in how you prompt, but it's pretty clear where this is going. The questions can stay simple while the answers get more sophisticated. The answers will in fact exceed the ability of the one asking.

Right now, competent people become very, very good with AI. They multiply themselves. The gap widens. But it does raise a question: what happens when the tools get so good and so specialised that the barrier drops for everyone? Does the edge compress again?

As I was pondering this, I generated a dashboard showing real-time top-performing ads directly from the Meta Ad Library across a competitor set. It took a few minutes. If you'd like the prompt, drop me a message and I'll send it over.