What differentiates an “AI-Native” architecture from simply “bolting on” an LLM API?

Written by

in