What are the biggest mistakes with productivity in General Local AI?

The point about mistakes and productivity is interesting. In practice I would keep the first test simple, record what worked, and only add extra complexity if it solves a specific limitation. The original point — "There are several ways to approach General Local AI, so I wanted to compare experiences. I am particularly interested in productivity and common mistakes. I would like to compare…" — gives a useful starting point for that comparison.