Somewhere on a dairy farm in Wellsville, Utah, are three cyborg  cows, indistinguishable from the rest of the herd.

Just like the other cows, they eat, drink, and chew their cud. Occasionally, they walk over to a big, spinning red-and-black brush, suspended at bovine back height, for a scratch. But while the rest of the cows just get their scratch and move on, these cows deliver data. Trackers implanted in their bodies use low-energy Bluetooth to ping a nearby base station and transfer information about the cows' chewing frequency, temperature, and general rambling around the farm.

These cows are the first to try a device called EmbediVet, created by a startup named Livestock Labs. For now, they're just going about their normal lives, unintentionally providing data that helps train an artificial neural network. The hope is that in the near future, this AI will help farmers figure out quickly and easily how well cows and other livestock are eating, whether they're getting sick or about to give birth—things that are typically done today just by watching and waiting but are difficult to spot when you've got hundreds or thousands of animals to keep an eye on.