How to automate anything
I’ve been thinking a lot lately about the relationship between manual tasks and automation. In this process, I’ve found myself describing how to move from one to the other in various contexts, and a pretty clear pattern has popped out at me. Here I want to distill the pattern down to the steps necessary to automate practically any process.
-
Document all the steps.
If the process is baking cookies, your list might look something like:
- Mix ingredients
- Form small balls from the dough, and place them on a baking sheet
- Place the baking sheet in the oven for a while
- Remove the baking sheet, and eat cookies
-
Create a checklist.
When baking cookies, this is usually called a recipe. This is a more detailed list of your list of steps. It should be sufficiently detailed that “anyone” can follow the instructions, and have the predictable desired result.
- Mix 2 cups flour, 1 cup sugar, 2 eggs…
- Preheat the oven
- Bake for 20 minutes
I like to think of this step as a sort of ”manual automation”. It’s designed to be done by a person, but in a way that requires no human ingenuity. In this case, the person acts as a human Turing machine, simply executing each step in the checklist.
-
Automate the single most painful task.
It’s usually a bad idea to automate an entire process all at once. There’s usually one constraint that’s acting as a bottleneck to your process. Identify the single most painful (or time-consuming, or costly) step, and automate it. Don’t worry about the other steps.
When baking cookies, it’s common to automate the “mixing” step, with some sort of electric mixer.
-
Observe, iterate.
After the single most painful task is automated, you may decide you’re done with your automation task. Unless you’re an elf living in a tree, it’s likely not worth the cost of automating the insertion and removal of cookies into the oven.
Or maybe you’re like my wife, and decide to purchase a kichen robot that automates many other parts of the cooking process (it measures, chops, mixes, even cooks some things).
If you find that your process is still painful, or later becomes painful (due to a change in context or scale), then return to step 3.
This is the same process I use when automating a build, or deployment, or setting up a CI/CD pipeline. If you’re interested, I discuss this in the context of CI/CD more at LeanCD.com.