The checklist test
One question, most of the answer
Owners often ask which of their tasks can be automated, expecting a technical answer about tools and models. The honest answer is a writing exercise. Imagine a competent temp starting Monday. Could you hand them a checklist for the task, and would following it produce an acceptable result?
If yes, the task is automatable today. If the checklist keeps sprouting it depends and just ask me, the task is not ready, and no tool will change that.
Why this test works
Modern AI is very good at following well-described procedures over messy inputs, and very bad at guessing what you meant but never said. The checklist test measures exactly that boundary. It is not about how complicated the task looks. Some intimidating tasks are long but perfectly mechanical. Some tiny tasks hide a judgment call in step two.
The test also surfaces the real blocker early. When owners cannot finish the checklist, it is usually because the task lives entirely in one person's head. That is not an automation problem, it is a key-person risk problem wearing a different hat.
Running the test on your week
Take your recurring tasks and sort them three ways:
- Checklist writes itself: automate these first
- Checklist mostly writes itself, with one or two judgment steps: automate around the judgment, leave a human at those steps
- Cannot write the checklist: document the task first, automate later, and notice you just found your riskiest process
The judgment steps are the design
The middle category is where good automation design lives. The machine does steps one through six, a person makes the call at step seven, the machine finishes eight through ten. Keeping the human at exactly the judgment step, and nowhere else, is most of the craft. It is also the shape of nearly everything we run for clients on retainer.
The audit answers this for your business
Two weeks, $2,500 flat ($1,000 for the first three clients), and you get the map of your own automatable work with dollars on it.