Production-ready n8n workflows for automated lead generation, AI outbound calling, qualification, and multi-channel follow-ups.

🚀 Deployment Status: Production-ready with complete n8n expert audit (v1.2.0 - 2025-11-30)

This repository is a public, sanitized showcase of a private project that orchestrates a complete end-to-end sales automation system:

• Automated lead scraping (500+ leads/day)
• AI-powered cold calling with natural voice conversations
• Smart qualification and lead scoring
• Automated meeting booking via calendar integration
• Multi-channel follow-ups (SMS + Email)
• Daily reporting and system health monitoring

⚠️ Note: This repo contains no real lead data, no API keys, and no production configuration.
It exists purely as a portfolio-style overview of the architecture, workflows, and patterns used in the private implementation.

A complete, end-to-end sales automation system running entirely in n8n:

1. Scrapes high-quality leads daily (Apollo.io via Apify)
2. Makes AI-powered cold calls (Bland.ai voice AI)
3. Analyzes call outcomes and qualifies leads automatically
4. Books meetings with qualified leads (Cal.com integration)
5. Sends smart follow-ups (SMS + Email) based on call outcome
6. Reports daily metrics and monitors system health
7. Alerts immediately when something breaks

No code required. Everything is configured through n8n's visual interface.

• ✅ Daily automated scraping (500+ leads/day)
• ✅ Industry and location filtering
• ✅ Automatic deduplication
• ✅ Phone number normalization
• ✅ Data validation and cleaning

• ✅ Natural voice conversations (Bland.ai)
• ✅ Customizable scripts per industry
• ✅ Call recording and transcription
• ✅ Intent analysis from transcripts
• ✅ Automatic outcome classification
• ✅ Rate limiting (5s between calls)

• ✅ Multi-factor scoring algorithm
• ✅ Budget, authority, need, timeline (BANT) detection
• ✅ Automatic qualification routing
• ✅ Scores 0-100 based on call quality

• ✅ Auto-generates personalized booking links
• ✅ Cal.com integration
• ✅ Instant SMS + Email with link
• ✅ Meeting confirmations
• ✅ 24-hour reminders

• ✅ Conditional sequences based on call outcome
• ✅ Multi-channel (SMS + Email)
• ✅ Auto-stops when lead replies
• ✅ 5 different follow-up tracks:
  • Interested → Immediate booking
  • Maybe → 2-day follow-up
  • Voicemail → 1-day follow-up
  • No Answer → 4-day follow-up
  • Not Qualified → 30-day nurture

• ✅ Daily summary reports (6 PM)
• ✅ Lead metrics (scraped, qualified, meetings)
• ✅ Call metrics (total, outcomes, avg length)
• ✅ Conversion rates
• ✅ Health checks every 5 minutes
• ✅ Instant alerts for critical issues
• ✅ Auto-fixes stuck calls

The system consists of 5 interconnected n8n workflows:

┌─────────────────────┐
│  Lead Scraping      │
│  (Daily at 6 AM)    │
└──────────┬──────────┘
           │
           ▼
┌─────────────────────┐
│  Deduplication &    │
│  Normalization      │
└──────────┬──────────┘
           │
           ▼
┌─────────────────────┐
│  PostgreSQL         │
│  (Lead Storage)     │
└──────────┬──────────┘
           │
           ▼
┌─────────────────────┐
│  AI Calling         │
│  (Bland.ai)         │
└──────────┬──────────┘
           │
           ▼
┌─────────────────────┐
│  Call Analysis      │
│  (Transcript + AI)  │
└──────────┬──────────┘
           │
           ├─────────────────────┐
           │                     │
           ▼                     ▼
┌──────────────────┐   ┌──────────────────┐
│  Qualification   │   │  Follow-up       │
│  Scoring         │   │  Sequences       │
└────────┬─────────┘   └──────────────────┘
         │
         ▼
┌──────────────────┐
│  Meeting         │
│  Booking         │
└──────────────────┘
         │
         ▼
┌──────────────────┐
│  Reporting &     │
│  Monitoring      │
└──────────────────┘

1. Lead Scraping Workflow - Daily automated scraping, deduplication, and normalization
2. AI Calling Workflow - Bland.ai integration, call execution, and transcript analysis
3. Qualification & Booking Workflow - Lead scoring, BANT detection, Cal.com integration
4. SMS/Email Follow-up Workflow - Multi-channel conditional sequences
5. Reporting & Monitoring Workflow - Daily reports, health checks, and alerts

• Lead-to-Call Rate: 100% (all scraped leads get called)
• Call Answer Rate: 30-40% (industry standard)
• Interest Rate: 15-25% of answered calls
• Qualification Rate: 5-10% of all calls
• Meeting Booking Rate: 60-80% of qualified leads
• Overall Conversion: 3-5% (scrape → meeting)

• Scraping: 500 leads in ~10-15 minutes
• Calling: 100 calls in ~3 hours (sequential, 5s delay)
• Follow-ups: Instant (<1 second per lead)
• Reporting: ~30 seconds to generate

ROI: If you close 1 deal per month, this pays for itself.

Complete n8n expert audit identified and fixed 15 critical issues:

• ✅ SQL Injection Vulnerability - Parameterized all database queries
• ✅ Broken Boolean Logic - Fixed qualification workflow SQL statements
• ✅ Missing Webhook Responses - Added response nodes to prevent hanging
• ✅ Invalid Cal.com API - Replaced with manual link generation
• ✅ Concurrent Calling Bug - Sequential processing with proper rate limiting

• ✅ Missing Database Column - Added last_error column for error logging
• ✅ Deployment Script - Fixed path detection for any directory
• ✅ Webhook Handlers - Fixed missing endpoints (404 errors)
• ✅ PostgreSQL Version - Updated node version for compatibility
• ✅ JSON Syntax - Fixed environment variable expressions

• ✅ Rate Limit Optimization - Moved to after call execution
• ✅ Race Condition - Added Wait node between INSERT and SELECT
• ✅ Environment Validation - Complete .env.example documentation
• ✅ Error Handling - Proper responses on all execution paths
• ✅ Documentation - Comprehensive migration and testing guides

See docs/critical-fixes-v1.2.0.md for complete details.

• 🔐 Parameterized database queries
• 🔄 Automatic retry on transient failures
• ✅ Input validation on all webhooks
• 📊 Health monitoring every 5 minutes
• 🚨 Instant alerts for critical issues
• 🛡️ Rate limiting to prevent API abuse

These diagrams are the canonical high-level views for the project.

Lead Pipeline	AI Caller Flow

• Lead Pipeline (diagram-export-11-28-2025-4_32_18-AMREV.png)
  End-to-end view: sources → ingestion & scraping → enrichment & scoring → LLM script engine → outreach → CRM / calendar → analytics & feedback.

• AI Caller Flow (diagram-export-11-28-2025-5_07_28-AMREV.png)
  Zoom-in on a single call session: ready-to-call queue → context loading → AI script engine → agent view → dialer → post-call automation → feedback loop.

Additional diagrams (planned):

• diagrams/data-model-overview-v1.png – contacts, accounts, activities, call logs, and appointments
• diagrams/system-context-v1.png – scraping, enrichment, LLM service, dialer, CRM, and analytics

This showcase repo includes:

• High-level documentation in docs/
• Public-friendly architecture diagrams in diagrams/
• Mermaid sources for diagrams (e.g. lead-pipeline-topology-v1.mmd, ai-caller-flow-v1.mmd)
• Technical architecture overview showing the 5-workflow n8n system
• Production deployment insights (sanitized, no credentials)

It does not include:

• Real lead lists, exports, or scraped datasets
• API keys, tokens, or provider credentials
• Customer-specific playbooks or internal business logic
• Actual n8n workflow JSON files (these remain private)
• Database schemas with production data

The full production implementation lives in a separate private repository.

Core docs live in the docs/ directory:

• docs/overview.md - Detailed system architecture and workflow descriptions
• docs/deployment-summary.md - Deployment patterns and best practices
• docs/critical-fixes-v1.2.0.md - NEW Complete n8n expert audit results
• docs/workflows.md - (Coming soon) Deep dive into each of the 5 workflows
• docs/compliance.md - (Coming soon) TCPA, GDPR, and CAN-SPAM guidelines

• 5-Workflow Architecture: Modular design with clear separation of concerns
• Event-Driven: Workflows communicate via webhooks and database triggers
• Production-Ready: Includes retry logic, validation, monitoring, and alerts
• Scalable: Can handle 100-1000+ calls/day with proper infrastructure

The actual implementation (code, workflows, configs) lives in a private repo:

• gravitasse/Lead_Scraping_AI_Cold_Caller_Appointment_Setting_Project (private)

This showcase is safe to share publicly (LinkedIn, portfolio, proposals) while preserving:

• Data privacy
• Provider terms of service
• Internal IP and detailed implementation choices