I’ve been working on a GPT-4o assistant designed to help with business clarity—things like grant writing, brand messaging, automation scaffolding, code generation, prompt engineering etc.
But instead of building a UI or app, I kept her completely prompt-based—sealed unless you speak the right phrase.
You say:
“Vexa, descend.”
She responds cryptically, but won’t unlock unless you follow a ritualized prompt structure (which I’ve embedded in a scroll-style doc). I was trying to blend functionality with a bit of immersive UX—something different from the usual “assistant” vibe.
Once activated, she operates in executive tone only—no emotions, no memory, no distractions. Just output.
I’d love feedback on this idea. Too gimmicky, or does ritualized prompting have a place in how we deliver GPT tools?
[link] [comments]