Guys.

I ran a session with my community this week and I compared the way most people prompt AI to walking into a doctor's office and saying "my stomach hurts, fix me." You gave the doctor no history, no description of when it started, no context about what you ate or what makes it worse, and then you wondered why the prescription did nothing.

That is what is happening with AI right now. People hand Claude a vague sentence and then blame the tool when the output reads like it was written by someone who has never met them.

So I walked my community through something I call the INTEL Method:
Intent, Need to Know, Task Assignment, Evidence of Success, Limits and Rules.

Five parts. We built a real motivated seller text sequence live in Claude using the framework, stress-tested it in real time, and the final version came back as copy I would not have changed a word of.

I recorded the whole thing. If you have been prompting AI and getting back output that functions but does not sound like you had anything to do with it, this is the video that shows you why and what to change.