This Espresso Machine Knows Exactly When You Need a Caffeine Shot

When you need caffeine, you need it right away. When that post-lunch drowsiness hits you, you don't have time to wait for a pick-me-up. Espressos are a great way to get your caffeine fix, but it takes time to make a proper espresso shot. The espresso machine has to heat up to the correct temperature before you can pull a shot. And, ideally, the machine should be at temperature for about 30 minutes first. That's too long to wait, which is why Mark Smith built an espresso machine that knows when he's ready for a latte.

Smith started with a Rancilio Silvia, which is midrange espresso machine with a built-in milk steamer. Before his modifications, Smith had to go through several steps to make a latte. First, he would start heating the water. Then, 30 minutes later, he would pull the espresso shot. After that, the water had to heat up to steam temperature, at which point he could steam the milk. Now, after the modifications, Smith can walk over to the machine and it will be hot and ready to go. After pulling the espresso shot, the machine will automatically heat up for steaming and let Smith know when it is ready.

This all happens thanks to the magic of machine learning. A machine learning model running on a Raspberry Pi Zero W monitors Smith's consumption happens and adjusts itself over time, similar to how Nest thermostats adapt to your habits. It can, for example, determine that Smith wants a latte around 2pm every weekday, but not on weekends. With that information, it can heat up the espresso machine 30 minutes before it knows that Smith will want a latte. Of course, Smith can also activate it other times if he's feeling unusually sleepy.

The Raspberry Pi monitors the espresso machine's water level through a float sensor and the water temperature through a thermocouple. A "bar graph" Nixie tube displays relevant data, such as the current water temperature in relation to the desired temperature. More granular data is available through a web interface hosted on the Raspberry Pi. All the components tuck away neatly inside of the espresso machine's enclosure in a 3D-printed case, so the machine looks unmodified aside from the Nixie tube.

If you want to build your own predictive espresso machine, Smith published all of the necessary files on his website.