This year I'm seeing folks fed up with instability in #AI-forward #cloud platforms and #SaaS, demanding that they slow down on feature delivery to focus on quality and reliability. That's heartwarming to me as a technologist, but it makes me wonder if AI can even meaningfully help?

We've seen that LLM codegen can shorten timelines for sloppy, okay-enough software. If customer demand turns towards reliability, can you just tune the prompts? Or does it basically invalidate the current LLM stack?