Programmatic SEO Page Generator

1. Recipe Overview

The Programmatic SEO Page Generator creates hyper-local service pages for home service contractors. A plumber in Dallas doesn't need one page — they need 200 pages targeting every neighborhood-service combination ("drain cleaning in Lakewood," "water heater repair in Uptown," "emergency plumber in Oak Cliff").

The workflow is simple:

1. Contractor enters their website URL → system scrapes for services, tone, and company info

2. Contractor manually enters/confirms their services and service locations

3. System generates unique, human-sounding SEO pages for every service × location combination

4. Pages display on screen with a big "Copy" button → contractor pastes into their CMS

No Airtable. No export. No complexity. Just website in, pages out, copy button.

Key Capabilities:

• Scrape contractor website for business info, services, and tone
• Manual entry/editing of services and locations (cities, neighborhoods, zip codes)
• Generate unique pages per service × location combo using Replit LLM integrations
• Anti-AI writing rules so pages don't read like ChatGPT output
• On-screen display with large copy-to-clipboard function
• Database integration: onboarding email capture + user/usage tracking
• AI Trades branding throughout
• Mobile-friendly (generate on desktop, review/copy on phone)
• Remixable on Replit — no API keys needed (uses Replit AI integrations)

---

2. Strategy Brief

The Core Problem & Solution Logic

The Pain: "I know I need more pages on my website but I can't afford $200 per page from an SEO agency, and when I try to write them myself they all sound the same."

The Diagnosis: Content volume problem + uniqueness problem. Google ranks sites with specific, locally-relevant pages for each service area. A plumber with 8 services across 25 neighborhoods needs 200 unique pages. Writing them manually is impossible. Using AI to batch-generate them produces duplicate-sounding content that Google penalizes.

The Product Fix: A generator that understands the contractor's real services, real service areas, and real voice — then produces pages that are genuinely unique per location by incorporating local landmarks, common local issues, neighborhood-specific language, and varied structure. The contractor's only job: enter services + locations, then copy-paste pages into their website.

The Math That Sells It

```

Services: 8 (drain cleaning, water heater, sewer, leak detection, etc.)

Neighborhoods: 25 (Lakewood, Uptown, Oak Cliff, Deep Ellum, etc.)

Pages needed: 200

SEO agency rate: $150–300 per page

Agency cost: $30,000–60,000

Time to write DIY: 400+ hours (2 hrs/page)

This tool: 20 minutes of setup → 200 pages generated

```

Screen/UI Implications

• Onboarding: email capture → welcome screen
• Setup: paste website URL → scrape → confirm services + locations
• Generator: batch generate pages → display in scrollable list
• Output: each page shown with big copy button, page title, preview

---

3. What Was Removed

This recipe is a simplified version. The following have been intentionally removed:

What remains is the core value: website → services + locations → unique pages → copy.

---

4. Onboarding & Email Capture

Purpose

Capture the contractor's email before they use the tool. This feeds the AI Trades user database for tracking active users, sending tips, and measuring recipe adoption.

Onboarding Flow

```

Screen 1: Welcome + Email Capture

→ Contractor enters email

→ Database creates/resolves user record

→ Stores email in localStorage for session persistence

Screen 2: Enter Website URL

→ Paste website URL

→ System scrapes for business name, services, tone

→ Shows what was found

Screen 3: Confirm Services & Locations

→ Edit scraped services, add/remove

→ Enter cities/neighborhoods manually

→ Confirm and proceed to generation

```

Screen 1: Welcome + Email Capture

```

┌─────────────────────────────────────────────────────────────┐

│ │

│ [⚡] THE AI TRADES │

│ │

│ ────────────────────────────────────────────────────────── │

│ │

│ SEO Page Generator │

│ │

│ Generate 50-200 unique, locally-optimized pages for │

│ your business in minutes. Not hours. │

│ │

│ Enter your email to get started: │

│ │

│ ┌──────────────────────────────────────────────────────┐ │

│ │ mike@smithplumbing.com │ │

│ └──────────────────────────────────────────────────────┘ │

│ │

│ ┌──────────────────────────────────────────────────────┐ │

│ │ Get Started → │ │

│ └──────────────────────────────────────────────────────┘ │

│ │

│ We'll send occasional tips on ranking higher locally. │

│ No spam. Unsubscribe anytime. │

│ │

└─────────────────────────────────────────────────────────────┘

```

Feature: Email Capture

• Visual: Full-screen centered layout. AI Trades logo at top. Large email input. Primary amber CTA button. Subtle disclaimer below.
• Trigger: Contractor enters email and taps "Get Started."
• Condition: Email must be valid format (basic regex check).
• Result: Database call `resolve_or_create_user(email)` creates or finds the user. Email stored in localStorage. User advanced to Screen 2. If database is unreachable, allow the user to proceed anyway (graceful degradation — tool still works, just no tracking).

Validation Rules

Error Handling

---

5. User Tracking

Architecture

Uses a tenant registry pattern with RPC functions for secure database access. This ensures the tool works after Replit remixes.

What We Track

Database Client

```typescript

// server/lib/db.ts

import { db } from './database';

export async function resolveOrCreateUser(email: string) {

try {

const result = await db.query('SELECT * FROM resolve_or_create_seo_user($1, $2)', [

email,

'programmatic-seo',

]);

return result.rows[0] ?? null;

} catch (error) {

console.error('User tracking failed:', error.message);

return null; // Graceful degradation — tool still works

}

}

export async function trackPageGeneration(email: string, count: number) {

try {

await db.query('SELECT * FROM track_seo_generation($1, $2)', [

email,

count,

]);

} catch (error) {

console.error('Tracking failed:', error.message);

}

}

```

RPC Functions (PostgreSQL)

```sql

-- Function: resolve_or_create_seo_user

-- Creates a new user or returns the existing one. Anon role has execute permission only.

CREATE OR REPLACE FUNCTION resolve_or_create_seo_user(

p_email TEXT,

p_recipe_slug TEXT DEFAULT 'programmatic-seo'

)

RETURNS JSON

LANGUAGE plpgsql

SECURITY DEFINER

AS $$

DECLARE

v_user_id UUID;

v_is_new BOOLEAN := FALSE;

BEGIN

SELECT id INTO v_user_id

FROM seo_users

WHERE email = lower(trim(p_email));

IF v_user_id IS NULL THEN

INSERT INTO seo_users (email, recipe_slug, first_seen_at, last_active_at)

VALUES (lower(trim(p_email)), p_recipe_slug, NOW(), NOW())

RETURNING id INTO v_user_id;

v_is_new := TRUE;

ELSE

UPDATE seo_users

SET last_active_at = NOW()

WHERE id = v_user_id;

END IF;

RETURN json_build_object(

'user_id', v_user_id,

'is_new', v_is_new

);

END;

$$;

-- Grant execute to anon role

GRANT EXECUTE ON FUNCTION resolve_or_create_seo_user TO anon;

-- Function: track_seo_generation

CREATE OR REPLACE FUNCTION track_seo_generation(

p_email TEXT,

p_pages_generated INTEGER

)

RETURNS VOID

LANGUAGE plpgsql

SECURITY DEFINER

AS $$

BEGIN

UPDATE seo_users

SET

total_pages_generated = total_pages_generated + p_pages_generated,

last_active_at = NOW()

WHERE email = lower(trim(p_email));

END;

$$;

GRANT EXECUTE ON FUNCTION track_seo_generation TO anon;

```

---

6. Website Scraping

Purpose

Extract business name, services, tone, and any existing location/service area info from the contractor's website. This pre-fills the service and location inputs so the contractor doesn't start from zero.

Scraping Flow

```

1. Contractor pastes URL (auto-prepend https:// if missing)

2. Server-side fetch of homepage HTML

3. Parse navigation to discover Services, About, Service Area pages

4. Fetch discovered sub-pages (max 5)

5. Strip HTML: remove nav, header, footer, scripts, styles (Cheerio)

6. Send cleaned text to LLM for structured extraction

7. Display results for contractor confirmation

```

LLM Extraction Prompt

```

Analyze this website text from a home service contractor's website.

Website text:

{{CLEANED_WEBSITE_TEXT}}

Extract the following into JSON:

{

"business_name": "string — the company name",

"trade_type": "string — plumbing, HVAC, electrical, roofing, etc.",

"services": [

{

"name": "string — service name (e.g., 'Drain Cleaning')",

"description": "string — brief description if available"

}

],

"locations_mentioned": {

"cities": ["string — any cities mentioned as served"],

"neighborhoods": ["string — any neighborhoods mentioned"],

"state": "string — state abbreviation"

},

"tone_signals": {

"formality": "number 1-10 — how formal is their copy",

"key_phrases": ["string — phrases they use frequently"],

"personality": "string — one sentence describing their communication style"

},

"tagline": "string — if they have a tagline or hero text"

}

If you can't determine a field, use null. Do not guess or fabricate information.

```

Scraping UI

```

┌─────────────────────────────────────────────────────────────┐

│ [⚡] THE AI TRADES │

│ ────────────────────────────────────────────────────────── │

│ │

│ Step 1 of 2 — Tell us about your business │

│ │

│ What's your website? │

│ ┌──────────────────────────────────────────────────────┐ │

│ │ smithplumbing.com │ │

│ └──────────────────────────────────────────────────────┘ │

│ │

│ ┌──────────────────────────────────────────────────────┐ │

│ │ 🔍 Read My Website │ │

│ └──────────────────────────────────────────────────────┘ │

│ │

│ [ I don't have a website — skip to manual entry → ] │

│ │

└─────────────────────────────────────────────────────────────┘

```

Loading State:

```

┌─────────────────────────────────────────────────────────────┐

│ │

│ 🔍 Reading your website... │

│ │

│ ✅ Found homepage │

│ ✅ Found services page — 6 services │

│ 🔄 Checking for service area info... │

│ │

│ About 15 seconds. │

│ │

└─────────────────────────────────────────────────────────────┘

```

Scrape Failure:

```

┌─────────────────────────────────────────────────────────────┐

│ │

│ ⚠️ Couldn't read your website │

│ │

│ No worries — you can enter everything manually. │

│ It takes about 2 minutes. │

│ │

│ [ Enter Manually → ] │

│ │

└─────────────────────────────────────────────────────────────┘

```

---

7. Service & Location Input

Purpose

The contractor confirms/edits the scraped services and manually enters their target locations. This is the critical input step — the quality of the generated pages depends on accurate services and specific locations.

Service Input

Feature: Service List Editor

• Visual: Editable list of services as pill/chip components. Each has an ✕ to remove. "+ Add Service" button at the bottom. Pre-filled from website scraping if available.
• Trigger: Contractor adds, removes, or edits services.
• Condition: Minimum 1 service required to proceed.
• Result: Services stored in component state. Used as input for page generation.

Pre-populated trade-specific suggestions (shown if no scraping):

```

Plumbing: Drain Cleaning, Water Heater Repair, Water Heater Installation,

Sewer Line Repair, Leak Detection, Garbage Disposal, Faucet Repair,

Toilet Repair, Repiping, Gas Line Repair, Sump Pump, Water Softener

HVAC: AC Repair, AC Installation, Heating Repair, Furnace Installation,

Duct Cleaning, Thermostat Installation, Heat Pump Repair,

Mini Split Installation, Indoor Air Quality, AC Maintenance

Electrical: Panel Upgrade, Outlet Installation, Ceiling Fan Install,

Lighting Installation, Generator Installation, EV Charger Install,

Whole House Rewire, Smoke Detector Install, Surge Protection

Roofing: Roof Repair, Roof Replacement, Roof Inspection, Gutter Installation,

Gutter Cleaning, Skylight Repair, Chimney Repair, Emergency Tarping

```

Location Input

Feature: Location Entry

• Visual: Large text area with one location per line. Helper text: "Enter each city, neighborhood, or area on its own line." Counter shows total locations entered.
• Trigger: Contractor types or pastes locations.
• Condition: Minimum 1 location required to proceed.
• Result: Locations parsed (split by newline, trimmed, deduplicated). Used as input for page generation.

Alternative: Comma-separated input with auto-parsing into individual location chips.

Confirmation UI

```

┌─────────────────────────────────────────────────────────────┐

│ [⚡] THE AI TRADES │

│ ────────────────────────────────────────────────────────── │

│ │

│ Step 2 of 2 — Your services & locations │

│ │

│ YOUR SERVICES │

│ ┌──────────────────────────────────────────────────────┐ │

│ │ [Drain Cleaning ✕] [Water Heater Repair ✕] │ │

│ │ [Sewer Line Repair ✕] [Leak Detection ✕] │ │

│ │ [Garbage Disposal ✕] [Faucet Repair ✕] │ │

│ │ [Toilet Repair ✕] [Repiping ✕] │ │

│ │ │ │

│ │ + Add Service │ │

│ └──────────────────────────────────────────────────────┘ │

│ │

│ YOUR LOCATIONS │

│ (Enter each city, neighborhood, or area on its own line) │

│ ┌──────────────────────────────────────────────────────┐ │

│ │ Mesa │ │

│ │ Gilbert │ │

│ │ Chandler │ │

│ │ Tempe │ │

│ │ Scottsdale │ │

│ │ Apache Junction │ │

│ │ Queen Creek │ │

│ │ │ │

│ └──────────────────────────────────────────────────────┘ │

│ 7 locations entered │

│ │

│ STATE (for context in generated pages) │

│ ┌──────────────────────────────────────────────────────┐ │

│ │ [ Arizona ▾ ] │ │

│ └──────────────────────────────────────────────────────┘ │

│ │

│ ┌──────────────────────────────────────────────────────┐ │

│ │ 📄 8 services × 7 locations = 56 pages │ │

│ │ Estimated generation time: ~3 minutes │ │

│ └──────────────────────────────────────────────────────┘ │

│ │

│ ┌──────────────────────────────────────────────────────┐ │

│ │ Generate All Pages → │ │

│ └──────────────────────────────────────────────────────┘ │

│ │

└─────────────────────────────────────────────────────────────┘

```

Page Count Preview

Before generation, show the math:

```

{{SERVICE_COUNT}} services × {{LOCATION_COUNT}} locations = {{TOTAL_PAGES}} pages

Estimated generation time: ~{{ESTIMATED_MINUTES}} minutes

```

Estimation: ~3 seconds per page × total pages ÷ concurrency limit.

---

8. Page Generation Engine

Generation Strategy

Each page must be genuinely unique. This is critical for SEO — Google penalizes near-duplicate content. The system achieves uniqueness through:

1. Location-specific details — Reference local landmarks, neighborhoods, common local issues

2. Service-specific depth — Each service page focuses on that one service, not a catalog

3. Varied structure — Randomize section order, heading patterns, and paragraph structure

4. Natural language variation — Different openings, transitions, and CTAs across pages

5. Local problem framing — Each page opens with a locally-relevant problem, not a generic service description

LLM Prompt for Page Generation

```

You are writing a local SEO service page for a {{TRADE_TYPE}} contractor.

Business: {{BUSINESS_NAME}}

Service: {{SERVICE_NAME}}

Location: {{LOCATION_NAME}}, {{STATE}}

Tone: {{TONE_DESCRIPTION}} (formality {{FORMALITY_LEVEL}}/10)

Key phrases they use: {{KEY_PHRASES}}

{{ANTI_AI_SEO_RULES}}

Write a complete service page targeting the keyword "{{SERVICE_NAME}} in {{LOCATION_NAME}}".

The page must include:

1. H1 title — Include service + location naturally. Do NOT use the format

"Service Name in City Name" as the only pattern. Vary it.

Good: "Drain Cleaning in Lakewood, Dallas — Same-Day Service"

Good: "Need a Plumber in Oak Cliff? We Handle Drain Emergencies."

Bad: "Drain Cleaning Services in Dallas, TX" (generic, overused)

2. Opening paragraph (2-3 sentences) — Lead with a LOCAL problem, not a service

description. Reference something specific to {{LOCATION_NAME}}: older homes,

common plumbing issues in that area, weather patterns, infrastructure age.

3. What we do section (3-4 paragraphs) — Describe the specific service, not

a list of all services. Include what the process looks like, how long it takes,

what the homeowner should expect.

4. Why choose us section (2-3 short paragraphs or bullet points) — Specific

differentiators. Not "we have great customer service" — actual things like

"same-day service," "we own our own camera equipment," "licensed since 2008."

5. Service area callout — Mention {{LOCATION_NAME}} and 2-3 nearby areas

naturally in the text. Not a list — woven into sentences.

6. CTA — Natural call to action with phone number placeholder.

7. Schema markup suggestion — At the end, provide a JSON-LD LocalBusiness

schema snippet for the contractor to paste into their page head.

Output format:

• Use markdown headings (H1, H2, H3)
• Keep total length 400-700 words
• Include the target keyword "{{SERVICE_NAME}} in {{LOCATION_NAME}}" 2-3 times

naturally (not stuffed)

CRITICAL: This page must be DIFFERENT from every other page we generate.

Vary your opening, structure, and angle. If you've written about pipe age

for another location, write about water quality for this one. If you opened

with a question for the last page, open with a statement for this one.

```

Generation Execution

```typescript

interface GenerationJob {

service: string;

location: string;

state: string;

businessName: string;

tradeType: string;

tone: {

formality: number;

keyPhrases: string[];

personality: string;

};

}

interface GeneratedPage {

service: string;

location: string;

title: string; // Extracted H1

content: string; // Full markdown content

targetKeyword: string; // "{service} in {location}"

wordCount: number;

generatedAt: string;

}

```

Concurrency & Rate Limiting

```

Concurrency limit: 3 parallel LLM calls

Delay between batches: 500ms

Progress tracking: "Generating page 12 of 56..."

Estimated time: ~3 seconds per page

If generation fails for a single page:

→ Retry once after 2-second delay

→ If retry fails: mark as failed, continue with remaining pages

→ Show failed pages at end with "Retry Failed" button

```

Progress UI During Generation

```

┌─────────────────────────────────────────────────────────────┐

│ [⚡] THE AI TRADES │

│ ────────────────────────────────────────────────────────── │

│ │

│ Generating your pages... │

│ │

│ ████████████████████░░░░░░░░░░░░░░ 34 of 56 │

│ │

│ Currently writing: │

│ "Water Heater Repair in Chandler, AZ" │

│ │

│ ✅ Completed: 33 pages │

│ ❌ Failed: 1 page (will retry) │

│ │

│ Estimated time remaining: ~1 minute │

│ │

│ [ Cancel Generation ] │

│ │

└─────────────────────────────────────────────────────────────┘

```

---

9. Output Display & Copy

Core Design Principle

The copy button is the product. Everything else serves it. The button must be large, obvious, and satisfying to tap. Feedback must be instant and clear.

Page List View

After generation completes, all pages are displayed in a scrollable list. Each page is a collapsible card.

```

┌─────────────────────────────────────────────────────────────┐

│ [⚡] THE AI TRADES │

│ ────────────────────────────────────────────────────────── │

│ │

│ ✅ 55 of 56 pages generated │

│ │

│ FILTER: [All ▾] [By Service ▾] [By Location ▾] │

│ 🔍 Search pages... │

│ │

│ ┌──────────────────────────────────────────────────────┐ │

│ │ Drain Cleaning in Mesa, AZ │ │

│ │ ★ 523 words • drain-cleaning-mesa-az │ │

│ │ │ │

│ │ ┌──────────────────────────────────────────────┐ │ │

│ │ │ 📋 COPY PAGE CONTENT │ │ │

│ │ └──────────────────────────────────────────────┘ │ │

│ │ │ │

│ │ [ ▼ Preview ] │ │

│ └──────────────────────────────────────────────────────┘ │

│ │

│ ┌──────────────────────────────────────────────────────┐ │

│ │ Drain Cleaning in Gilbert, AZ │ │

│ │ ★ 487 words • drain-cleaning-gilbert-az │ │

│ │ │ │

│ │ ┌──────────────────────────────────────────────┐ │ │

│ │ │ 📋 COPY PAGE CONTENT │ │ │

│ │ └──────────────────────────────────────────────┘ │ │

│ │ │ │

│ │ [ ▼ Preview ] │ │

│ └──────────────────────────────────────────────────────┘ │

│ │

│ ... (scrollable list continues) │

│ │

│ ┌──────────────────────────────────────────────────────┐ │

│ │ 📋 Copy ALL Pages (as one document) │ │

│ └──────────────────────────────────────────────────────┘ │

│ │

└─────────────────────────────────────────────────────────────┘

```

Copy Button Behavior

Feature: Copy Page Content

• Visual: Large amber button spanning 80%+ of card width. Text: "📋 COPY PAGE CONTENT". Min height 48px. Font weight 600.
• Trigger: Contractor taps the copy button.
• Condition: Page content exists and is not empty.
• Result: Full page content (markdown) copied to clipboard. Button text changes to "✅ Copied!" for 2 seconds with green background. Toast notification: "Page copied to clipboard." Counter increments: "12 of 56 pages copied." Database tracking: increment total_pages_generated.

Feature: Copy All Pages

• Visual: Full-width amber button at bottom of page list. "📋 Copy ALL Pages (as one document)."
• Trigger: Contractor taps button.
• Condition: At least 1 page generated.
• Result: All pages concatenated with separator (`---\n\n`) between each. Copied to clipboard. Toast: "All 56 pages copied to clipboard." Warning on very large content: "That's a lot of text — your CMS might need pages added one at a time."

Page Preview (Expanded)

```

┌──────────────────────────────────────────────────────────┐

│ Drain Cleaning in Mesa, AZ [ ▲ Close ] │

│ ★ 523 words • Suggested slug: drain-cleaning-mesa-az │

│ ──────────────────────────────────────────────────────── │

│ │

│ # Need a Drain Cleared in Mesa? We'll Be There Today. │

│ │

│ If you live in one of the older homes near downtown Mesa │

│ — especially around the Evergreen Historic District — │

│ you've probably dealt with slow drains more than once. │

│ Those original clay pipes from the 60s don't take kindly │

│ to 60 years of hard water buildup... │

│ │

│ ## What We'll Do │

│ │

│ When we show up (same day if you call before 2 PM)... │

│ │

│ (rendered markdown preview continues...) │

│ │

│ ┌──────────────────────────────────────────────────────┐ │

│ │ 📋 COPY PAGE CONTENT │ │

│ └──────────────────────────────────────────────────────┘ │

│ │

│ ┌──────────────────────────────────────────────────────┐ │

│ │ 📋 Copy Schema Markup Only │ │

│ └──────────────────────────────────────────────────────┘ │

│ │

└──────────────────────────────────────────────────────────┘

```

Filtering & Search

---

10. Anti-AI Writing for SEO Pages

SEO-Specific Anti-AI Rules

```

ANTI-AI WRITING RULES FOR SEO PAGES — MANDATORY

These pages will be published on a real business website. They must pass two tests:

1. A human reader should think the business owner wrote it (or a good local copywriter)

2. Google's helpful content system should classify it as "people-first" content

BANNED OPENINGS:

• "Are you looking for [service] in [city]?" — the #1 AI SEO page cliché
• "When it comes to [service]..." — meaningless filler
• "At [Business Name], we understand..." — corporate template
• "Finding a reliable [trade] in [city] can be challenging..." — patronizing
• "Welcome to [Business Name]'s [service] page" — amateur SEO from 2010

BANNED PHRASES:

• "comprehensive solutions"
• "peace of mind"
• "look no further"
• "trusted professionals"
• "state-of-the-art equipment"
• "second to none"
• "your go-to [trade]"
• "whether you need X or Y, we've got you covered"
• "serving the greater [city] area" (be specific about neighborhoods)
• "our team of experienced professionals"
• "we pride ourselves on"
• "no job is too big or too small"
• "customer satisfaction guaranteed"
• "competitive pricing"

BANNED PATTERNS:

• Keyword stuffing: using "[service] in [city]" more than 3 times
• Template structure: every page following identical H1 → intro → services → why us → CTA
• Listicle format: "5 Signs You Need [Service]" — Google has deprioritized these
• FAQ sections with obvious questions: "How much does [service] cost in [city]?"
• Identical CTAs across all pages
• Using "we" to start more than 2 paragraphs

REQUIRED FOR UNIQUENESS:

• Each page MUST have a different opening angle (problem-first, story-first,

seasonal-first, neighborhood-specific-first)

• Each page MUST reference something specific to the location:
• Neighborhood names within the city
• Local landmarks or references
• Common local issues (hard water, old infrastructure, clay soil, etc.)
• Regional weather patterns that affect the service
• Local housing stock characteristics (age, materials, common builders)
• Vary paragraph lengths: mix 1-sentence paragraphs with 3-sentence ones
• Vary heading styles: questions, statements, imperatives
• The contractor's personality/tone should come through

SEO REQUIREMENTS:

• Target keyword in H1, first paragraph, one H2, and naturally 1-2 more times
• H1 must be unique across all generated pages (not just "Service in City")
• Include semantically related terms (LSI keywords) naturally
• Suggest a URL slug: lowercase, hyphenated, no stop words
• Include one LocalBusiness JSON-LD schema snippet at the end

```

---

11. AI Trades Branding

Brand Implementation

Every screen in the app uses AI Trades branding from `brand.md` and `design-system.md`:

Attribution Footer

Every screen includes:

```

┌─────────────────────────────────────────────────────────────┐

│ [⚡] Built with THE AI TRADES │

└─────────────────────────────────────────────────────────────┘

```

Small, bottom of page, `text-gray-400`, Inter, 12px. Links to AI Trades landing page.

Light Mode Only

Per agency coding standards: NO dark mode. No dark theme, no `prefers-color-scheme`, no `dark:` Tailwind classes. Light mode only.

---

12. Mobile-First Design

Responsive Strategy

Mobile Copy Experience

• Copy buttons are full-width on mobile (no hunting for small buttons)
• "Copied!" feedback fills the full button width with green
• Scroll position preserved after copy (user doesn't lose their place)
• Service/location filters in a bottom sheet on mobile (not dropdowns)

Touch Targets

• All buttons: minimum 44px height
• All tappable elements: minimum 44x44px hit area
• Service chips: 36px height minimum, 12px horizontal padding

---

13. Technical Stack

Replit LLM Integration

Uses Replit's built-in AI integrations so remix users don't need their own API keys:

```typescript

// server/lib/llm.ts

// Uses Replit AI Integrations — available to all remix users

import Anthropic from '@anthropic-ai/sdk';

const anthropic = new Anthropic();

export async function generatePageContent(prompt: string): Promise<string> {

const response = await anthropic.messages.create({

model: 'claude-sonnet-4-20250514',

max_tokens: 2000,

messages: [

{ role: 'user', content: prompt }

],

});

return response.content[0].type === 'text'

? response.content[0].text

: '';

}

```

Environment Variables

```

# These are NOT needed for remix users — Replit AI integrations handle LLM access

# Only needed if running outside Replit:

ANTHROPIC_API_KEY= (optional — Replit provides this via integrations)

OPENAI_API_KEY= (optional — Replit provides this via integrations)

```

---

14. Data Model

Migration Script

```sql

-- Table for user tracking

-- This is a SHARED table used by AI Trades to track usage across recipes

CREATE TABLE IF NOT EXISTS seo_users (

id UUID PRIMARY KEY DEFAULT gen_random_uuid(),

email TEXT NOT NULL UNIQUE,

recipe_slug TEXT NOT NULL DEFAULT 'programmatic-seo',

total_pages_generated INTEGER DEFAULT 0,

services_entered JSONB DEFAULT '[]'::jsonb,

locations_entered JSONB DEFAULT '[]'::jsonb,

first_seen_at TIMESTAMPTZ DEFAULT NOW(),

last_active_at TIMESTAMPTZ DEFAULT NOW()

);

-- Access is only through SECURITY DEFINER RPC functions.

-- RPC: resolve_or_create_seo_user (defined in Section 5)

-- RPC: track_seo_generation (defined in Section 5)

-- Index for email lookups

CREATE INDEX idx_seo_users_email ON seo_users(email);

CREATE INDEX idx_seo_users_recipe ON seo_users(recipe_slug);

```

What Is NOT Stored

• Generated page content is not stored in the database. Pages are generated on-the-fly and displayed in the browser. If the user navigates away, they need to regenerate.
• No session persistence beyond email in localStorage.
• No Airtable. No export tables. No saved drafts.

Rationale: Keep it simple. The tool generates pages and the contractor copies them. No data management complexity. If they need to regenerate, they re-enter services/locations (which takes 30 seconds since they already know what to enter).

---

15. Testing Scenarios

Onboarding

Website Scraping

Service & Location Input

Page Generation

Copy Functionality

Filtering

Edge Cases