01 / Inventory the work
List ten tasks you repeat. Note how often each occurs and what makes it slow.
02 / Score the candidates
For each task, consider repetition, time spent, sensitivity, error impact, and ease of human review. Avoid high-risk tasks as your first experiment.
03 / Select a small pilot
Choose one task with clear inputs and measurable outcomes, such as structuring nonconfidential notes.
04 / Compare the result
Measure the time taken, accuracy, revisions needed, and whether the final result was genuinely better.
The idea, at a glance
01Task
02Frequency
03Risk
04Reviewability
05Pilot
Your next move: List five possible uses and circle the safest, easiest one to test.
Explore more free resources