How we tracked the Three Kings across Extremadura thanks to AI

Anyone who knows me knows I am a journalist and that my coding experience was below zero, so when Arturo Chaves Pecero challenged me to build a web tool tracking the Three Kings across Extremadura, the sensible move would have been to commission it from an outside supplier. I opened ChatGPT.

The challenge

In a digital newsroom, ideas for interactive formats nearly always die at the same point. Something occurs to you for one specific night, you add up what it costs to commission outside and how much time is left before 5 January, and the idea goes back in the drawer until the following year. What we wanted here was not something we had: a live piece you could look up, answering the only question that matters that night, where are the Kings right now, and holding up at ten at night and at one in the morning without anybody touching it.

What we did

A week of long conversations with the model, a lot of prompting and a fair number of steps backwards, until what we were after finally turned up, all of it from inside the building:

• An HTML and CSS front end, built for mobile, which is where it was going to be seen.
• A JavaScript back end interpolating the coordinates between one point and the next.
• A JSON file with the full route and a timestamp on every coordinate, which the AI also built for me after a great deal of prompting.
• A counter of presents delivered that climbed along with the caravan and which, honestly, was the part people liked most.

Thanks to those timestamps, anyone connecting at any hour always saw the exact point the caravan had reached, without anyone in the newsroom having to touch a thing that night.

The outcome

More than 16,000 people followed the live route in the first edition, and in the second the counter reached 27,325 presents handed out across the region. We ran it again in January 2025 with the piece already road-tested.

The interesting part, beyond the number, is what it opens up behind it: a digital newsroom can try a new format in a week, with the people it already has, and get it wrong cheaply when it does not work. We still have plenty of ideas in that drawer, and there are far fewer excuses now for leaving them there.