A person with a list as the first matching engine is good advice, and it's cheap to try. I'd add one habit: write down every manual decision the human makes in the first two weeks, including the ones that feel obvious. Those notes end up being the spec for the automated version, and the odd exceptions are where the first algorithm would have gone wrong.
StackShortlist
Proof Over Promises.
A person with a list as the first matching engine is good advice, and it's cheap to try. I'd add one habit: write down every manual decision the human makes in the first two weeks, including the ones that feel obvious. Those notes end up being the spec for the automated version, and the odd exceptions are where the first algorithm would have gone wrong.