Watching everyone get excited about GPT-5's prompt optimizer got me thinking about how this whole thing evolved. I have been letting AI write its own prompts for at least two years. Started as my default approach when I realised something obvious: if the AI understands what I want better than I can explain it, why spend time crafting perfect instructions?
Apparently this is now considered innovative.
How This Actually Started
Back when everyone was learning STAR frameworks and role-based prompting, I got lazy. Instead of crafting elaborate instructions, I started saying: "Figure out how to handle this. Design your own process."
Turned out the AI created better approaches than my careful prompt engineering ever did.
The Mindset Difference
"I need to learn how to talk to AI."
"AI needs to learn how to talk to my business problems."
Different approach. Different results.
"They're optimizing instructions. I'm optimizing outcomes."
Instead of building prompt libraries, I just tell AI what outcomes I need and let it figure out the systematic approach. I never write prompts for operations, workflows, or customer service systems anymore. The AI consistently creates more thorough approaches than manual prompt engineering ever produced.
What Changes When You Think This Way
You stop thinking about AI as a tool that follows instructions. You start thinking about AI as a business partner that develops its own methodologies. The AI does not just create better outputs — it creates systems that get smarter over time.
If You Want to Try This
Three shifts to make
The AI is smart enough to figure out its own methodology if you are clear about what success looks like. The next level is AI that understands business context well enough to anticipate problems before you describe them. Systems that develop institutional knowledge and adapt accordingly. That is what treating AI as a strategic partner looks like.
Anyone else notice how techniques you have been using for a while suddenly become the "next big thing"?