ROUTEDevtools & code
Jev Testing Learnings
Ben Dus@beeendus
This project details learnings from testing an AI router within a coding workflow, revealing initial evaluation flaws.
1/14
An AI router got 24/24 decisions right in my test.
Then I put it inside a coding workflow. It showed no completion advantage.
Its first apparent win turned out to be a flaw in my evaluation.
Here's what I learned testing Jev.
Related builds
