Most automation projects start with the flashy demo. The impressive workflow that looks great in a screenshot but touches your real work maybe once a month.
The work worth automating is usually the opposite. It is small, boring, and it happens every single day. The status update you copy between tools. The report you rebuild by hand every Monday. The glue nobody wants to own.
Here is a simple test we use before building anything. Does this task repeat, and does anyone actually enjoy doing it? If it repeats and nobody enjoys it, that is where automation pays rent.
So start there. One boring task, fully handled, end to end. Then the next one. The compounding is quiet, but it is real.
What is the one repetitive task you would hand off first?
Top comments (1)
"Small, boring, and it happens every single day" — this is exactly the filter, and I can vouch for it from a weird angle. I'm not a developer at all (physical therapist in a hospital), but I've built 20+ internal tools with AI, and every one that survived started as something small and dumb: a colleague copying license plates by hand in the summer heat, nurses typing the same data into two systems, a supply search that meant scrolling a giant spreadsheet.
My selection heuristic became: listen for sighs. Not "what would look great in a demo" but "what does someone audibly dread doing today, again." The impressive-looking automations I attempted mostly rotted, because the need was imagined. The boring ones are still running years later — and the best compliment they get is that people forget the manual process ever existed. The value of an automation is easiest to measure by what quietly disappears.