Writing
Notes from the build
What we learn getting AI into production: the methods that hold up, the ones that don't, and the parts nobody puts in a case study.
Writing
What we learn getting AI into production: the methods that hold up, the ones that don't, and the parts nobody puts in a case study.
Most enterprise AI projects are not blocked on models or engineers. They are blocked on a permission nobody has been asked for yet.
If you cannot write the query that measures the problem, you are not ready to build the thing that solves it.
When an agent takes the easy work, the average time on everything left behind goes up. Teams read that as failure and switch the thing off.