That one sentence is the entire principle. This product is the operating manual for it.
Not a template pack. A method for building prompts grounded in your specific business — so the output sounds like you, knows your constraints, and solves your actual problem.
No subscription. No upsell. No fake scarcity.
Marcus runs a 3-person web design studio. He quoted a local bakery on a $3,500 website two weeks ago. They were warm on the call, then stopped responding. Here is what happens when he prompts without context — and what happens when he applies the Architecture.
Marcus / Follow-up email — Web design studio, Portland. Proposal out to Sugarcoat Bakery, $3,500 site. Two weeks of silence. He needs to follow up without sounding needy or desperate.
Write a follow-up email for a client who hasn't responded to my proposal.
Subject: Following up on my proposal Hi [Client Name], I wanted to follow up on the website proposal I sent over. I'm still very interested in working with you and would love to help bring your vision to life! Please let me know if you have any questions or if you'd like to schedule a call to discuss. I'm flexible with timing and happy to adjust the proposal if needed. Looking forward to hearing from you! Best, [Your Name]
Business context: I run a small web design studio (3 people) focused on local service businesses — bakeries, cafes, boutiques. Clients hire us because we get local and don't treat them like a number. Role: Write as me. I'm direct and low-pressure. I don't chase clients — I make it easy for them to say yes or no. Instruction: Follow-up email to Sugarcoat Bakery on a $3,500 proposal. Sent two weeks ago. They were warm on the call, then went quiet. Constraints: Don't apologize for following up. No "just checking in." Under 100 words. One ask only. Format: Email with subject line.
Subject: Sugarcoat — quick note Hi Maya, Wanted to check whether the timing still works on your end. We have a slot open in our June build schedule that would fit a launch before summer. If the budget or scope needs adjusting, I'm happy to do a quick call. If the timing is not right, just say the word — no pressure either way. Marcus
The first prompt asked for an email. The second gave the model a business type, a person, a situation, and a constraint set — so it could produce something that sounds like Marcus, not a template. The AI did not get smarter. The model is the same. The question is the same. The context is the entire difference. That is the principle. The Architecture is how you apply it consistently, to every problem, in every session.
The Marcus scenario is one of five. Each one covers a different high-stakes situation, with the same four-part structure: real setup, weak-but-plausible prompt, context-loaded prompt, and both AI outputs side by side. Hard-won contrast, not manufactured demos.
Not a collection of prompts. A complete method for getting consistently useful output from AI — grounded in your specific business, not generic best practices.
The full cheat sheet is 12 rows. Here are three. You will leave this page knowing something concrete — that is intentional. If the free rows are useful, the full sheet is inside the pack along with the method that generates all of them.
| When you're writing… | Give AI… | Skip… |
|---|---|---|
| A follow-up email e.g. proposal, intro, check-in | Your business type, the relationship stage, what "no response" usually means in your context, one constraint on tone | "Write a professional follow-up email" — generic framing produces generic output every time |
| Positioning copy e.g. homepage, offer page, bio | The one audience you are actually trying to reach, their specific objection, and what makes you different from the obvious alternatives | "Describe what I do" — without the audience's objection loaded, AI writes for everyone and converts no one |
| A stuck decision e.g. pricing, hire vs. contract, pivot | Both options spelled out, what each costs if you get it wrong, and the one constraint that actually can't move | "Help me decide between A and B" — AI gives you a balanced pros/cons list when you need a recommendation grounded in your actual situation |
| + 9 more rows in the full cheat sheet → Get the pack for $69 | ||
Knowledge is free. The paid pack is the complete system — method, worked examples, cheat sheet, all five scenarios.
Each card shows a real SMB situation — the vague prompt most owners type, and the B.R.I.E.F. version that produces something actually useful. The model is identical in both columns. The difference is what you give it.
The generic prompt gets a bullet list of specs. The B.R.I.E.F. prompt gets a description that leads with the right buyer's actual motivation — upgrading family, turnkey kitchen — and the edge cases field silently handles two real liability risks (roof disclosure, fair housing language) without requiring the agent to remember them mid-session.
The vague prompt produces a five-section proposal that could have been written for any client. The B.R.I.E.F. version knows the real sales situation — competing against larger agencies, a client burned by "deliverables" language, and a rebrand they're still sensitive about. The edge cases field catches the landmines before they're stepped on.
Without context, AI writes a description for a mass-market mug — "stylish," "perfect for on-the-go." With context, it writes for someone who specifically chose a $58 handmade mug over a $20 one. The edge cases field prevents two real problems: a vague insulation claim that causes returns, and missing the hand-wash caveat that will generate support emails.
The vague prompt gets a narrative summary of what the client said. The B.R.I.E.F. version produces a structured memo that flags the asset transfer as time-sensitive and separates legally relevant facts from the client's interpretation — because the edge cases field explicitly prevents editorializing and surfaces what the reviewing attorney actually needs to act on.
Without context, AI writes a hype post that sounds like every other food brand on Instagram. With context, it writes something that sounds like the actual restaurant — the local farm sourcing, the family story behind the dish, the anti-hype voice. The edge cases field ("don't say 'limited time only' like a Starbucks launch") is the line between generic food content and something regulars will actually respond to.
One-time purchase. Instant download. Works with ChatGPT, Claude, Gemini, or any AI tool that accepts text input.
No subscription. No upsell. No fake scarcity.
No subscription. No upsell. No fake scarcity.