[AIorNot]

One easy way to judge the quality of of the spec the ai generates is to run it a few times on the same story and compare the differences

Curious if you tried that - how much variation does the AI do or does the grounding in codebase and prompts keep it focused and real?

[antonybrahin]

I haven't done intense tests yet, but based on my preliminary tests, the output is about 80% consistent. The others are like suggesting additional changes.