Lesson 8 of 8100% complete

20 min read

Real Claygent Examples

Complete prompt workflows for common use cases

Why These Examples Are Different

These aren't "find company info" examples. They're battle-tested workflows from real GTM engineering projects that achieve 85%+ accuracy. Each example solves a problem that native integrations can't handle well.

ℹ️

These prompts use techniques like URL subdomain detection (95% accuracy vs 70% for text scraping) and multi-stage waterfalls with confidence stacking.

How AutoClaygent Handles This

AutoClaygent generates prompts like these automatically. Describe what you need in plain English and it produces the prompt, schema, and tests it until it scores 8.0+.

Example 1: Platform Detection via URL Patterns

Use case: Detect what SaaS platforms a company uses by analyzing their portal URLs and login pages.

Why it's valuable: URL patterns are 95%+ accurate. Companies use vendor-hosted portals that reveal their tech stack through subdomain patterns.

The Prompt

Given the company domain: {{domain}}

Detect what platforms this company uses by analyzing their portal/login URLs.

**Research Steps:**

1. Look for customer-facing portals:
   - Check {{domain}}/login, {{domain}}/portal, {{domain}}/app
   - Look for "Client Login", "Patient Portal", "Customer Portal" links
   - Check the footer for portal links

2. When you find a portal link, analyze the URL pattern:

   **Subdomain Patterns (highest confidence):**
   - {company}.platformname.com → Uses Platform Name
   - app.{company}.com redirecting to vendor → Uses that vendor
   - portal.vendorname.com/{company} → Uses Vendor Name

   **Common Platform URL Patterns:**
   - *.salesforce.com → Salesforce
   - *.hubspot.com → HubSpot
   - *.zendesk.com → Zendesk
   - *.calendly.com → Calendly
   - *.webflow.io → Webflow
   - *.mindbody.io → Mindbody
   - *.jane.app → Jane App

3. Check for scheduling/booking tools:
   - Look for "Book Now", "Schedule", "Appointments"
   - These often reveal Calendly, Acuity, Cal.com, etc.

**Output as JSON:**
{
  "platforms_detected": [
    {
      "platform_name": "Name of platform",
      "platform_category": "CRM | Support | Scheduling | Chat | Website",
      "detection_method": "subdomain | redirect | widget | integration_page",
      "evidence_url": "URL that revealed this"
    }
  ],
  "portal_url": "Customer portal URL if found" | null,
  "confidence": "high" | "medium" | "low"
}

The JSON Schema

{
  "type": "object",
  "properties": {
    "platforms_detected": {
      "type": "array",
      "items": {
        "type": "object",
        "properties": {
          "platform_name": {"type": "string"},
          "platform_category": {
            "type": "string",
            "enum": ["CRM", "Support", "Scheduling", "Chat", "Website", "EHR", "Billing"]
          },
          "detection_method": {
            "type": "string",
            "enum": ["subdomain", "redirect", "widget", "integration_page"]
          },
          "evidence_url": {"type": "string"}
        },
        "required": ["platform_name", "platform_category", "detection_method", "evidence_url"]
      }
    },
    "portal_url": {"anyOf": [{"type": "string"}, {"type": "null"}]},
    "confidence": {"type": "string", "enum": ["high", "medium", "low"]}
  },
  "required": ["platforms_detected", "portal_url", "confidence"],
  "additionalProperties": false
}

Sample Output

{
  "platforms_detected": [
    {
      "platform_name": "Jane App",
      "platform_category": "EHR",
      "detection_method": "subdomain",
      "evidence_url": "https://downtownchiro.janeapp.com"
    },
    {
      "platform_name": "Calendly",
      "platform_category": "Scheduling",
      "detection_method": "widget",
      "evidence_url": "https://calendly.com/downtown-chiro/consult"
    }
  ],
  "portal_url": "https://downtownchiro.janeapp.com",
  "confidence": "high"
}

Why This Works

Detection Method	Accuracy	Why
Subdomain patterns	95%+	DNS doesn't lie
URL redirects	90%	Login flows reveal vendors
Widget detection	85%	Embedded scripts have signatures
Text scraping	70%	Companies don't list their tools

Example 2: SMB Owner Discovery Pipeline

Use case: Find the owner of a small business when Apollo and other databases come up empty.

Why it's valuable: No single source covers SMB. A 5-stage waterfall with confidence stacking achieves 76% coverage vs 15% from any single source.

The Prompt

Given the company:
- Name: {{company_name}}
- Domain: {{domain}}
- Address: {{address}}

Find the owner or primary decision-maker for this small business.

**5-Stage Waterfall (stop when confident):**

**Stage 1: Google Maps**
Search: "{{company_name}} {{address}}"
- If owner name listed → Extract it
- If only business name → Continue to Stage 2

**Stage 2: Website Team Page**
Check {{domain}}/about, {{domain}}/team, {{domain}}/our-story
- Look for "Owner", "Founder", "Principal"
- Check for single-person bios (common for SMB)

**Stage 3: Domain WHOIS**
Check WHOIS for {{domain}}
- Registrant name (if not privacy-protected)
- Administrative contact

**Stage 4: Cross-Reference Sources**
If you have a candidate name from any stage:
- Search: "{{name}}" "{{company_name}}" owner
- Look for LinkedIn confirmation
- Check local business registrations

**Stage 5: Web Search Fallback**
Search: "{{company_name}}" owner OR founder "{{city}}"
- Check news articles, awards, features
- Chamber of Commerce listings

**Confidence Stacking:**
- 1 source = low (50%)
- 2 sources agree = medium (75%)
- 3+ sources = high (90%+)

**Output as JSON:**
{
  "owner_name": "Full name" | null,
  "owner_title": "Owner | Founder | CEO | Principal" | null,
  "sources_found": ["google_maps", "website", "whois", "linkedin", "news"],
  "source_count": number,
  "evidence": "Brief description of where you found this",
  "confidence": "high" | "medium" | "low"
}

The JSON Schema

{
  "type": "object",
  "properties": {
    "owner_name": {"anyOf": [{"type": "string"}, {"type": "null"}]},
    "owner_title": {
      "anyOf": [
        {"type": "string", "enum": ["Owner", "Founder", "CEO", "Principal", "President", "Managing Partner"]},
        {"type": "null"}
      ]
    },
    "sources_found": {
      "type": "array",
      "items": {
        "type": "string",
        "enum": ["google_maps", "website", "whois", "linkedin", "news", "chamber", "yelp"]
      }
    },
    "source_count": {"type": "integer", "minimum": 0, "maximum": 7},
    "evidence": {"anyOf": [{"type": "string"}, {"type": "null"}]},
    "confidence": {"type": "string", "enum": ["high", "medium", "low"]}
  },
  "required": ["owner_name", "owner_title", "sources_found", "source_count", "evidence", "confidence"],
  "additionalProperties": false
}

Sample Output (High Confidence)

{
  "owner_name": "Dr. Michael Chen",
  "owner_title": "Owner",
  "sources_found": ["google_maps", "website", "linkedin"],
  "source_count": 3,
  "evidence": "Listed as owner on Google Maps, bio on /about page, confirmed on LinkedIn as 'Owner at Downtown Chiro'",
  "confidence": "high"
}

Sample Output (Medium Confidence)

{
  "owner_name": "Sarah Johnson",
  "owner_title": "Principal",
  "sources_found": ["website"],
  "source_count": 1,
  "evidence": "Only source on /our-story page mentioning 'Sarah Johnson, Principal'",
  "confidence": "medium"
}

💡

Confidence Stacking

The magic is in the source count. One source saying "John Smith is the owner" might be outdated. Three sources agreeing? That's reliable data.

More Examples in the Full Course

The complete course includes worked examples for all 9 patterns:

Example	Problem It Solves	Key Innovation
CRM Validation	Find WHERE your CRM data is wrong	Output match/mismatch/stale with evidence
Business Model Classification	Classify by revenue model	Decision tree beats "find pricing info"
Buying Intent Detection	Find companies ready to buy NOW	Multi-signal temporal analysis
Corporate Structure	Find actual decision-maker	Identify parent vs subsidiary

Key Takeaways

URL patterns beat text scraping — 95% accuracy vs 70%
Waterfalls beat single sources — 76% coverage vs 15%
Confidence stacking is essential — Count your sources
Decision trees beat generic prompts — Explicit logic wins
Return null when unsure — Wrong data is worse than no data

← PreviousQuality Scoring Complete! →Back to Home

Real Claygent Examples

Why These Examples Are Different

Example 1: Platform Detection via URL Patterns

The Prompt

The JSON Schema

Sample Output

Why This Works

Example 2: SMB Owner Discovery Pipeline

The Prompt

The JSON Schema

Sample Output (High Confidence)

Sample Output (Medium Confidence)

More Examples in the Full Course

Key Takeaways

Ready to build Claygents that actually work?