Prompt 2 - Intersection Design
Prompt 3 - Behavioral Role Programming
Prompt 4 - Role Validation Design
Something clicked for me when I was running my blockchain agency.
I'd been asking AI to "act as a business consultant" for months. Getting decent advice, nothing terrible. But then one day I tried something different.
Instead of "business consultant," I said "act as a blockchain startup advisor who's helped 20+ Web3 companies navigate their first token launch, specializing in regulatory compliance and community building strategies."
The response was completely different. Not just better - it was like talking to someone who actually understood the specific challenges of blockchain startups instead of generic business theory.
That's when I realized most people are activating maybe 10% of the AI's available expertise.
Here's what happens when you use generic roles:
"You are a marketing expert."
The AI defaults to the most common marketing advice patterns in its training data. Blog posts about "5 marketing tips." General principles that apply to everyone and therefore help no one specifically.
You get vanilla recommendations because "marketing expert" could mean anything. B2C? B2B? Email marketing? Brand strategy? Performance marketing? Content marketing?
The AI has to guess, so it plays it safe with broadly applicable advice that lacks the specificity you actually need.
Think about how knowledge actually works in the real world.
A "marketing expert" who specializes in SaaS growth hacking thinks completely differently from someone who does luxury brand storytelling. They've developed different mental models, different approaches, different frameworks.