Reusable Prompts Are More Leverage Than Reusable Code

Every Monday you assemble the same status digest by hand: skim three channels, pull the open items, decide which two actually matter, write it up. It takes twenty minutes. By the third week you think the thought every knowledge worker eventually thinks: I should automate this. I should write a script.

You sketch it on a napkin. You'd need API access to the channels, a parser, somewhere to run it, and a rule for "actually matters" that, you realize on the napkin, you can't quite write down. The script goes on the someday list, where it stays. Meanwhile you keep doing it by hand, twenty minutes a week, for a year.

The script was never the bottleneck. The fact that you'll never write it is.

#The script you'll never write

Code is the prestige answer to "automate this," so it's the one most people reach for. It's also the one with the highest activation cost. A script has to be specified, written, debugged, hosted, and maintained; it also has to be specified exactly, in advance, by someone who can write it.