Post

I worked in Japan. My co-worker and I didn't speak the same language. I was given data, we discussed requirements with an interpreter, I wrote a spec which was developed by someone in another country. The first test of the interface processed 800+ transactions and 3 failed. Common goals are good.

I'll use that in a company presentation ...

And that way lies my biggest concern with the current state of LLM based coding assistants: The extreme focus on automating the generation of the written representation.