Step 1: Finding High-Intent, Low-Tech Local Businesses
I open Google Maps targeting mid-sized US cities (Portland, Austin, Charlotte). I look specifically for businesses with 4.5+ stars on Google Reviews but either NO website link or a website built in 2011 that fails on mobile.
If a business has 120 positive reviews on Google, they have paying customers and cash flow. Their outdated website is actively costing them money every day.
Step 2: Structuring the AI Prompt Architecture
Instead of asking generic chatbots to "build me a website", I structure the prompt into 4 distinct phases: Wireframe, Copywriting, Tailwind Styling, and Call-to-Action placement.
By providing the business’s actual menu items, address, and high-resolution unsplash photography in the initial system prompt, the AI generates a prototype that feels personalized rather than a generic template.
System Prompt: "You are a principal design engineer specializing in high-converting local business landing pages. Create a Next.js component using Tailwind CSS for an artisanal sourdough bakery located in Portland, OR. Sections required: Sticky Hero with pickup hours, Seasonal Sourdough grid with dietary tags, Customer Testimonials with Google star ratings, and an interactive Location/Order modal."
Step 3: Deploying Live to a Free Link (The Proof of Work)
Never send a mock PDF or an image. Always send a live, clicking, responsive URL deployed to Vercel or Netlify. When the owner clicks a link on their iPhone and sees their own bakery name, hours, and photos loading in 400ms, the sale is 80% done.
Putting this into practice
The entire build took 2 hours and 40 minutes. The email outreach took 15 minutes. The client approved the invoice on Stripe 48 hours later. This is the exact playbook we execute together during Saturday and Sunday in The Weekend Build.