← Alle cases
Own project · AI showcase

Bietenradar: from office joke to national media in a few days

A local idea, built with AI: bietenradar.nl predicts, per neighbourhood in Groningen, how strong the sugar factory's beet smell is. From first idea to a working, live tool in a few days.

Screenshot of bietenradar.nl: the map with the odour plume and the per-neighbourhood prediction in Groningen.
Screenshot of bietenradar.nl: the map with the odour plume and the per-neighbourhood prediction in Groningen.
3 days
from idea to a working live tool
560
neighbourhoods with their own prediction
1 day
to national and regional media
24/7
a new prediction every hour
01 · De uitdaging

The trigger: a Groningen topic that returns every year

Every autumn the Cosun Beet Company factory at Vierverlaten in Hoogkerk processes the region's beets. For weeks the smell hangs over Groningen. Every year the same discussion, the same jokes and the same question: does it smell today? Nobody had an answer, because the smell had never been predicted per neighbourhood.

We wanted to show how quickly you can go from idea to a working tool with AI. A local, recognisable topic is the perfect shop window for that.

The trigger

Together with Claude we looked for a Groningen topic that returns every year. The beet smell was right there.

The problem

The smell had never been predicted per neighbourhood. 'Does it smell today?' went unanswered.

The goal

From idea to a working, live tool in a few days, as a showcase of how we build.

02 · Onze aanpak

The approach: from weather data to a per-neighbourhood smell forecast

We built a tool with AI that combines weather data every hour with an odour model and resident reports, and translates that into a single score per neighbourhood.

01

Idea found with Claude

Together with Claude we searched for seasonal topics that return every year in Groningen. The beet smell came out right away, and it was spot on.

02

Weather data every hour

Via Open-Meteo we pull the wind direction, wind speed, boundary-layer height and low cloud at the factory every hour.

03

Odour model per neighbourhood

From the factory we draw an odour plume in the wind direction, with an e-folding distance of 6 kilometres and a surcharge when the air stays put. Across the 560 CBS neighbourhoods (via PDOK) we calculate a score per neighbourhood from 0 (nothing) to 4 (windows shut).

04

Residents report along

Smell something? With one tap you report it for your own neighbourhood, so the prediction gets sharper every day.

03 · Het resultaat

Within a few days there was a working, live tool: bietenradar.nl, with a map of 560 neighbourhoods, a score per neighbourhood and resident reports. The prediction refreshes every hour based on the latest weather data.

Within a day of launch NPO Radio 1, RTV Noord, Dagblad van het Noorden, Sikkom, OOG Groningen and even TikTok picked it up. Exactly what a good shop window does: show how quickly you go from idea to working product with AI.

De les

A good idea plus AI is enough to put something working live in days. Technology is no longer the bottleneck. The question is what you make, and whether a human makes the calls.

From office joke to national media in a few days. That is exactly what we mean: with AI you turn an idea into something that really works at lightning speed, with a human making the calls.

Marc Cornelius
Founder, Empathree