Early on, the unscalable work is the work
A note on Do Things That Don't Scale by Paul Graham, published on paulgraham.com.
Written for startup founders, and the audience that most needs it is engineers who have become founders.
The argument is that the manual work you would never build a product around, recruiting users one at a time, doing support personally, handling things by hand, is not a temporary embarrassment before the real business starts. It is how you find out what the real business is.
The engineering-specific failure it addresses is automating too early. Building an onboarding flow before you have onboarded anyone by hand means encoding your guesses about what people need into software, then discovering the guesses were wrong, then having to unpick the software as well as the assumption. Doing it manually ten times is slower per instance and enormously faster overall.
For anyone selling their own product rather than only client work, the specific line worth internalising is about founders being unwilling to do the unscalable parts because they feel unglamorous. The unglamorous parts are where the information is.
It is fifteen minutes and it is one of the few pieces of startup writing that is about doing rather than about attitude.
This is the thinking behind why cost cutting is motion rather than a strategy.