Four days on whether a tool works for everyone, who should get the final say, what an answer really costs, and whose work a model used.
Test three tools on the groups they skipped.
Decide which tool decisions need a person.
Find out what each headline actually counted.