Post

Finally got it through my head that AI works better with a goal, a list of constraints, and a list of how to validate done. I had been telling it how to do things and found it often went off the rails badly. This has changed *everything* for me