- AI-Proof Businesses: "ChatGPT can’t fix your broken truck at 2 AM"
- The Arbitrage: "Your competitor’s website is from 2003. You can build better in 4 days."
- Real Problems = Real Money: "Emergency diesel repair = $1,500. SaaS trial = $0"
The hook: "While everyone’s building AI wrappers, we’re using AI to dominate businesses AI can’t replace"
From never coding to live website in 5 minutes
1. Claude Code (1 minute)
- Go to: claude.ai/download
- Download Claude for Mac/PC
- Open it, sign in with email
- That’s it – no terminal, no config files
2. GitHub Account (30 seconds)
- Go to: github.com
- Click "Sign up"
- Use email + password
- Verify email
- Done – free account works perfectly
3. Vercel Account (30 seconds)
- Go to: vercel.com
- Click "Sign up"
- Choose "Continue with GitHub" (easiest)
- Authorize connection
- Done – free tier is perfect for this
Open Claude Code and say:
Create a new Astro project for a diesel mechanic website
Claude will:
- Create the folder structure
- Install dependencies
- Set up Astro
- Show you it’s ready
Still in Claude Code, say:
Initialize git and create a new GitHub repository called diesel-dudes
First time setup dialog:
Claude: "I need to authenticate with GitHub. Can I open your browser?"
You: "Yes" (Browser opens, you click "Authorize")
Claude then runs:
git init
git add .
git commit -m "Initial commit"
gh repo create diesel-dudes --public --source=.
git push -u origin main
Say to Claude:
Deploy this to Vercel
First time Vercel setup:
- Claude: "Please log into Vercel in your browser"
- Vercel: "Import Git Repository"
- Click your "diesel-dudes" repo
- Click "Deploy"
- Wait 45 seconds...
- BOOM – Live website!
The Magic:
"I never typed a command. I never saw a terminal. I just talked to Claude."
From design to deployed website
- Show Figma: Original diesel mechanic design
- Anima Export: "One click to React components"
- The Problem: Show bloated Anima code
<div className="overlap-group-5" style={{width: '1440px'}}>
Say:
Convert this Anima code to optimized Astro components
Show the result:
<section class="hero">
<h1>24/7 Mobile Diesel Repair</h1>
<!-- Clean, semantic, optimized -->
</section>
- "Zero JavaScript by default"
- "Perfect Lighthouse scores out of the box"
- "Builds to static HTML = instant SEO advantage"
20 optimizations that destroy local competition
You can’t rank for keywords you haven’t found
The discovery process:
First, I used ChatGPT and Claude together:
ChatGPT: "Generate 50 keywords for a mobile diesel mechanic in Charlotte"
Claude: "Analyze these keywords by search intent and buying stage"
Show the keyword categories discovered:
- Emergency keywords: "truck broke down I-77", "24 hour diesel repair"
- Service keywords: "mobile brake repair", "onsite oil change"
- Problem keywords: "diesel won’t start", "black smoke from exhaust"
- Local keywords: "diesel mechanic charlotte", "truck repair columbia sc"
The keyword mapping:
- Homepage → "mobile diesel mechanic"
- Emergency page → "24 hour", "emergency", "urgent"
- Service pages → "[service] + mobile/onsite"
- Location pages → "diesel mechanic [city]"
- Blog → Question/problem keywords
The insight: "Most competitors target 'diesel repair.' We found 'mobile diesel mechanic' had less competition but high intent."
"Fix the invisible problems first"
Live demo:
Analyze this site for technical SEO issues
Show the problems:
- No robots.txt
- No sitemap
- Mixed www/non-www
- Missing meta descriptions
Fix them all:
Fix all technical SEO issues — create robots.txt, XML sitemap, and canonical URLs
Plain English: "This is like fixing the plumbing before painting the walls"
"Help Google understand exactly what you do"
Show the progression:
Show me our current schema markup
→ Generic LocalBusiness
Upgrade to AutoRepair schema with full service catalog
→ Specific VehicleRepairBusiness with 10 services
Visual comparison:
- Before: Google sees "some local business"
- After: Google sees "24/7 diesel repair with emergency service, 4.9 stars, serves Charlotte–Columbia"
The result: Rich snippets, better rankings, voice search ready
"Thin content doesn’t rank"
Location page transformation:
Expand this Charlotte location page to 1,200 words with local details
Show before/after:
- Before: "We serve Charlotte" (50 words)
- After: Neighborhoods, truck routes, industrial areas, weather impacts (1,200 words)
Service page expansion:
Expand brake repair service to 800 words focusing on our process and transparency
The multiplier effect: "Do this for 10 services × 10 locations = 100 ranking pages"
"Scale your SEO audit with parallel agents"
Live demo:
Launch three agents:
1. Find all missing alt text across the site
2. Identify pages under 500 words
3. Audit for duplicate meta descriptions
Explain the magic:
- "Each agent gets fresh context"
- "They work simultaneously"
- "I get synthesized results, not raw data"
Show results: "Agent found 31 images needing alt text, 6 thin pages, 3 duplicate metas"
"Speed = Rankings (especially mobile)"
The image optimization:
Convert all images to WebP and implement lazy loading
Show the impact:
- Before: 38MB of images, 4.2s load time
- After: 2MB of images, 1.1s load time
- "95% size reduction in 2 minutes"
Core Web Vitals:
Optimize for Core Web Vitals — extract critical CSS and eliminate render blocking
"Own your geographic area"
NAP consistency:
Standardize our business info across all pages to (803) 230-6390
Local content injection:
Add Charlotte-specific truck routes, industrial areas, and breakdown hotspots to relevant pages
The compound effect: "Every local signal stacks. We just added 50+ while competitors have 2"
"Spread the SEO juice"
Quick implementation:
"Create internal linking between all related services and locations"
Show the network effect: Service pages > Location pages > Blog posts > Homepage
Part 5: The Results (2 min)
"From invisible to dominant"
Live Rankings Check:
- "mobile diesel mechanic charlotte" > Page 1
- "emergency diesel repair |-77" > Position 3
- "diesel mechanic near me" > Maps pack
Performance Metrics:
-
95% image compression
-
73% CSS reduction
-
100 Lighthouse scores
-
4.9 schema rating displayed