Back in 2016, James Makinson and I used harmonic radar to track every move made by individual forager bumblebees throughout their entire lives. We argued that each trip outside the nest could be classed as either an exploration flight, in which they seem to mix searching for flowers with learning about their surroundings, or an exploitation flight in which they focussed on efficiently collecting pollen or nectar from previously discovered flower patches. In this clever work, Sam Williams and colleagues at Bangor University came up with an efficient way to automate the categorisation of each flight. The really clever bit is that the algorithm could categorise flights with high accuracy from just the first few datapoints in a radar track. In other words, they can predict what the purpose of a trip is before the bee has even got to the flowers. This opens up lots of possibilities for automatically monitoring bee activity on a far greater scale than we can currently do, and the Bangor team hope to deploy it on a fancy new drone-based bee tracker. I didn’t play a huge role in this one, mostly providing training data and a bit of advice.