FuseSell Local

Complete AI-powered sales automation pipeline that runs entirely on your local machine.

FuseSell Local is a production-ready implementation of the FuseSell AI sales automation system, converted from server-based YAML workflows to a comprehensive Python command-line tool with full data ownership and privacy control.

Latest release: fusesell==1.2.1 is available on PyPI via pip install fusesell.

Contributors should review the Repository Guidelines before opening a pull request.

🚀 Complete Pipeline Overview

FuseSell Local processes leads through a complete 5-stage AI-powered pipeline:

1. Data Acquisition ✅ - Multi-source customer data extraction (websites, business cards, social media)
2. Data Preparation ✅ - AI-powered customer profiling and pain point analysis
3. Lead Scoring ✅ - Advanced product-customer fit evaluation with detailed scoring
4. Initial Outreach ✅ - Intelligent email generation with multiple personalized approaches
5. Follow-up ✅ - Context-aware follow-up sequences with interaction history analysis

Status: 100% Complete - Production Ready

🚀 Quick Start

Installation

1. Install Python dependencies:

cd fusesell-local
pip install -r requirements.txt

That's it! No additional setup required. The system automatically creates all necessary database tables and default configurations on first run.

2. Test the installation:

python fusesell.py --openai-api-key YOUR_API_KEY \
                   --org-id test_org \
                   --org-name "Test Company" \
                   --input-website "https://example.com" \
                   --dry-run

First Run (Complete Pipeline)

# Full end-to-end sales automation
python fusesell.py \
  --openai-api-key "sk-your-key" \
  --org-id "mycompany" \
  --org-name "My Company Inc" \
  --input-website "https://targetcompany.com"

This single command will:

1. Extract customer data from the website
2. Analyze and structure the data using AI
3. Score the lead against your products
4. Generate personalized email drafts
5. Create follow-up sequences

Library Integration

FuseSell Local can now be imported directly so orchestrators like RealtimeX can execute the pipeline without spawning the CLI.

1. Install the package in your Python environment:

   pip install fusesell
   

2. Run the pipeline programmatically with default helpers:

   from fusesell_local import execute_pipeline
   
   result = execute_pipeline(
       {
           "openai_api_key": "sk-your-key",
           "org_id": "mycompany",
           "org_name": "My Company Inc",
           "full_input": "Seller: My Company Inc, Customer: Target Corp, Communication: English",
           "input_website": "https://targetcompany.com",
       }
   )
   
   print(result["status"])
   

3. For finer-grained control (custom logging, shared storage, etc.), compose the lower-level utilities:

   from fusesell_local import (
       build_config,
       configure_logging,
       prepare_data_directory,
       run_pipeline,
       validate_config,
   )
   
   options = {
       "openai_api_key": "sk-your-key",
       "org_id": "mycompany",
       "org_name": "My Company Inc",
       "full_input": "Seller: My Company Inc, Customer: Target Corp, Communication: English",
       "input_website": "https://targetcompany.com",
       "customer_timezone": "America/New_York",
   }
   
   config = build_config(options)
   prepare_data_directory(config)
   configure_logging(config)
   valid, errors = validate_config(config)
   if not valid:
       raise ValueError(errors)
   
   outputs = run_pipeline(config)
   

When embedding FuseSell inside ephemeral interpreter services, consider supplying a custom data_dir scoped per run and set auto_configure_logging=False if you prefer stdout logging.

4. Reuse the packaged CLI inside scripts or tests:

   from fusesell_local import FuseSellCLI
   
   cli = FuseSellCLI()
   cli.run([
       "--openai-api-key", "sk-your-key",
       "--org-id", "mycompany",
       "--org-name", "My Company Inc",
       "--full-input", "Seller: My Company Inc, Customer: Target Corp, Communication: English",
       "--input-description", "Example Corp lead from automation script",
       "--dry-run",
   ])
   

📋 Complete Usage Examples

Example 1: Full Pipeline - Website to Follow-up

# Complete end-to-end sales automation
python fusesell.py \
  --openai-api-key "sk-proj-abc123" \
  --org-id "mycompany" \
  --org-name "My Company Inc" \
  --input-website "https://targetcompany.com"

Complete Pipeline Execution:

1. Data Acquisition: Scrapes website, extracts company information
2. Data Preparation: AI analysis of company profile and pain points
3. Lead Scoring: Evaluates product-customer fit with detailed scoring
4. Initial Outreach: Generates 4 personalized email draft variations
5. Follow-up: Creates context-aware follow-up sequence strategies

Exa

mple 2: Multi-Source Data Collection

# Combine multiple data sources for comprehensive profiling
python fusesell.py \
  --openai-api-key "sk-proj-abc123" \
  --org-id "mycompany" \
  --org-name "My Company Inc" \
  --input-website "https://targetcompany.com" \
  --input-business-card "https://example.com/business-card.jpg" \
  --input-linkedin-url "https://linkedin.com/company/targetcompany" \
  --input-facebook-url "https://facebook.com/targetcompany" \
  --input-description "Leading fintech startup in NYC, 50+ employees"

Multi-Source Processing:

• Website: Company information, services, team details
• Business Card: Contact information via OCR processing
• LinkedIn: Professional information, company updates, connections
• Facebook: Business information, customer engagement, posts
• Description: Additional context and insights

Example 3: Email Generation and Management

# Generate initial outreach emails only
python fusesell.py \
  --openai-api-key "sk-proj-abc123" \
  --org-id "mycompany" \
  --org-name "My Company Inc" \
  --input-description "Customer: John Smith at Acme Corp, email: john@acme.com, industry: manufacturing" \
  --stop-after initial_outreach

Email Generation Features:

• 4 Draft Approaches: Professional Direct, Consultative, Industry Expert, Relationship Building
• Personalized Subject Lines: 4 variations per draft
• Personalization Scoring: 0-100 score based on customer data usage
• Draft Comparison: Side-by-side analysis with recommendations

Example 4: Email Sending with Smart Scheduling

# Send email with optimal timing
python fusesell.py \
  --action send \
  --selected-draft-id "draft_professional_direct_abc123" \
  --recipient-address "john@acme.com" \
  --recipient-name "John Smith" \
  --org-id "mycompany" \
  --customer-timezone "America/New_York" \
  --business-hours-start "09:00" \
  --business-hours-end "17:00"

Smart Scheduling Features:

• Timezone Intelligence: Respects customer's business hours
• Optimal Timing: 2-hour default delay with business hours respect
• Weekend Handling: Automatically schedules for next business day
• Database Events: Creates events for external app processing

Example 5: Follow-up Email Generation

# Generate context-aware follow-up emails
python fusesell.py \
  --action draft_write \
  --stage follow_up \
  --execution-id "exec_abc123_20241209" \
  --org-id "mycompany"

Follow-up Intelligence:

• Interaction Analysis: Analyzes previous email history and engagement
• Strategy Selection: Chooses appropriate follow-up approach (1st, 2nd, 3rd, final)
• Context Awareness: References previous interactions appropriately
• Respectful Limits: Maximum 5 follow-ups with graceful closure

Example 6: Draft Rewriting and Improvement

# Rewrite existing draft based on feedback
python fusesell.py \
  --action draft_rewrite \
  --selected-draft-id "draft_consultative_xyz789" \
  --reason "Make it more technical and focus on ROI benefits" \
  --org-id "mycompany"

Draft Rewriting Features:

• LLM-Powered Rewriting: Uses AI to incorporate feedback
• Version Control: Tracks all rewrites with history
• Personalization Maintenance: Keeps customer-specific details
• Improvement Tracking: Monitors changes and effectiveness

Example 7: Pipeline Control and Testing

# Stop after lead scoring (don't generate emails)
python fusesell.py \
  --openai-api-key "sk-proj-abc123" \
  --org-id "mycompany" \
  --org-name "My Company Inc" \
  --input-website "https://targetcorp.com" \
  --stop-after lead_scoring

# Skip follow-up generation
python fusesell.py \
  --openai-api-key "sk-proj-abc123" \
  --org-id "mycompany" \
  --org-name "My Company Inc" \
  --input-website "https://targetcorp.com" \
  --skip-stages follow_up

# Dry run (no API calls, uses mock data)
python fusesell.py \
  --openai-api-key "test-key" \
  --org-id "test" \
  --org-name "Test Company" \
  --input-website "https://example.com" \
  --dry-run

# Debug mode with detailed logging
python fusesell.py \
  --openai-api-key "sk-proj-abc123" \
  --org-id "mycompany" \
  --org-name "My Company Inc" \
  --input-website "https://targetcorp.com" \
  --log-level DEBUG

Example 8: Process Continuation

# First run - stop after data preparation
python fusesell.py \
  --openai-api-key "sk-proj-abc123" \
  --org-id "mycompany" \
  --org-name "My Company Inc" \
  --input-website "https://targetcorp.com" \
  --execution-id "lead_001" \
  --stop-after data_preparation

# Continue from where we left off
python fusesell.py \
  --openai-api-key "sk-proj-abc123" \
  --org-id "mycompany" \
  --org-name "My Company Inc" \
  --continue-execution "lead_001" \
  --continue-action "approve_and_continue"

📊 Command Reference

Required Arguments

Data Sources (At Least One Required)

Optional Context Fields

Pipeline Control

Output Options

Team & Project Settings

Advanced Options

Email Scheduling Options

Action-Based Operations

Stage-Specific Operations

🌐 Enha

nced Web Scraping with Serper API

Optional but Recommended: Add --serper-api-key for better data collection:

# Enhanced scraping capabilities
python fusesell.py \
  --openai-api-key "sk-proj-abc123" \
  --serper-api-key "your-serper-key" \
  --org-id "mycompany" \
  --org-name "My Company Inc" \
  --input-website "https://targetcompany.com"

Benefits:

• 🌐 Better Website Scraping: More reliable content extraction
• 🔍 Company Research: Automatic Google search for company info
• 📱 Social Media Access: Enhanced LinkedIn/Facebook scraping
• 🛑 Graceful Fallback: Works without Serper API (shows warnings)

Get Serper API Key: Visit serper.dev -> Sign up -> Get free API key

🔄 Complete Pipeline Stages

1. Data Acquisition ✅

Purpose: Multi-source customer data extraction

Data Sources:

• Website Scraping: Company information, contact details, business description
• Business Card OCR: Contact extraction from images (Tesseract, EasyOCR, Cloud APIs)
• LinkedIn Profiles: Professional information, company details, connections
• Facebook Pages: Business information, contact details, public posts
• Manual Input: Structured customer descriptions

AI Processing:

• LLM-powered information extraction and structuring
• Multi-source data merging and conflict resolution
• Company research via search APIs

Output: Comprehensive customer profile with contact information

2. Data Preparation ✅

Purpose: AI-powered customer profiling and analysis

AI Analysis:

• Customer Profiling: Structured company information extraction
• Pain Point Identification: Categorized business challenges and priorities
• Financial Analysis: Revenue estimation, growth potential, funding sources
• Technology Assessment: Digital maturity and technology stack analysis
• Competitive Analysis: Market positioning and competitive landscape
• Development Planning: Growth plans, timeline estimates, resource requirements

Output: Enriched customer profile with pain points and business insights

3. Lead Scoring ✅

Purpose: Advanced product-customer fit evaluation

Scoring Framework:

• 5 Weighted Criteria: Industry fit, company size, pain points, technology, budget
• ROI Analysis: Payback period estimation and financial impact assessment
• Implementation Feasibility: Technical complexity and resource requirements
• Competitive Positioning: Market advantage and differentiation analysis
• Feature Alignment: Product capabilities vs. customer needs matching
• Scalability Assessment: Growth potential and expansion opportunities

Output: Detailed scoring breakdown with product recommendations and justifications

4. Initial Outreach ✅

Purpose: Intelligent email generation and draft management

Email Generation:

• 4 Approach Variations: Professional Direct, Consultative, Industry Expert, Relationship Building
• Personalized Subject Lines: 4 variations per draft with company-specific messaging
• Advanced Personalization: 0-100 scoring based on customer data usage
• Call-to-Action Optimization: Automatic CTA extraction and optimization

Draft Management:

• Comparison System: Side-by-side draft analysis with recommendations
• Version Control: Track original drafts and all rewrites with history
• Selection Algorithms: Configurable criteria-based best draft selection
• Customer Readiness: Outreach readiness scoring with improvement recommendations

Output: Multiple personalized email drafts with management tools

5. Follow-up ✅

Purpose: Context-aware follow-up sequences with interaction analysis

Interaction Analysis:

• History Tracking: Days since last interaction, total attempts, engagement patterns
• Sentiment Detection: Customer response analysis and engagement level scoring
• Sequence Intelligence: Automatic progression through follow-up stages
• Respectful Limits: Maximum 5 follow-ups with graceful closure handling

Follow-up Strategies:

• Gentle Reminder (1st): Friendly check-in with soft approach
• Value-Add (2nd): Industry insights, resources, and helpful information
• Alternative Approach (3rd): Different angle, case studies, social proof
• Final Attempt (4th): Respectful closure with future opportunity maintenance
• Graceful Farewell (5th): Professional relationship preservation

Smart Features:

• Timing Intelligence: Minimum 3-day intervals between follow-ups
• Context Awareness: References previous interactions appropriately
• Engagement Adaptation: Adjusts tone and approach based on customer behavior

Output: Context-aware follow-up emails with sequence management

🔍 M

anaging Multiple Sales Processes

When running multiple sales processes, use the querying tools:

# List all recent sales processes
python query_sales_processes.py --list

# Get complete details for a specific process
python query_sales_processes.py --details "fusesell_20251010_141010_3fe0e655"

# Find processes by customer name
python query_sales_processes.py --customer "Target Corp"

# Get specific stage results
python query_sales_processes.py --stage-result "task_id" "lead_scoring"

📚 Complete querying guide: QUERYING_GUIDE.md

📁 Data Storage & Configuration

All data is stored locally in the fusesell_data directory with 100% server-compatible schema:

fusesell_data/
+--- fusesell.db              # SQLite database
+--- config/                  # Configuration files
|   +--- prompts.json        # LLM prompts
|   +--- scoring_criteria.json
|   +--- email_templates.json
+--- drafts/                 # Generated email drafts
+--- logs/                   # Execution logs

Database Tables

• executions - Execution records and metadata
• customers - Customer profiles and information
• lead_scores - Lead scoring results and breakdowns
• email_drafts - Generated email drafts and variations
• stage_results - Intermediate results from each stage

🔧 Server-Compatible Database Schema

FuseSell Local uses exact server table names for seamless integration:

• llm_worker_task: Task management (matches server exactly)
• llm_worker_operation: Stage execution tracking (matches server exactly)
• gs_customer_llmtask: Customer data storage (matches server exactly)
• executions: Backward compatibility VIEW that maps to llm_worker_task

Configuration Files

Custom Prompts (config/prompts.json)

Customize LLM prompts for different stages:

{
  "data_preparation": {
    "customer_analysis": "Your custom prompt for customer analysis...",
    "pain_point_identification": "Your custom prompt for pain points..."
  },
  "lead_scoring": {
    "product_evaluation": "Your custom scoring prompt..."
  }
}

Scoring Criteria (config/scoring_criteria.json)

Customize lead scoring weights and criteria:

{
  "criteria": {
    "industry_fit": { "weight": 25, "description": "Industry alignment" },
    "company_size": { "weight": 20, "description": "Company size fit" },
    "pain_points": { "weight": 30, "description": "Pain point match" },
    "technology": { "weight": 15, "description": "Technology compatibility" },
    "budget": { "weight": 10, "description": "Budget indicators" }
  }
}

🎯 Key Features

✅ Complete AI-Powered Sales Automation

• Multi-Source Data Collection: Websites, business cards (OCR), LinkedIn, Facebook
• AI Customer Profiling: Pain point analysis, company research, financial assessment
• Intelligent Lead Scoring: Product-customer fit evaluation with detailed breakdowns
• Personalized Email Generation: 4 different approaches with subject line variations
• Context-Aware Follow-ups: Smart sequence management with interaction history analysis

✅ 100% Local Execution & Privacy

• Complete Data Ownership: All customer data stays on your machine
• No External Dependencies: Except OpenAI API for LLM processing
• SQLite Database: Local data storage with full CRUD operations
• Event-Based Scheduling: Database events for external app integration
• Comprehensive Logging: Detailed execution tracking and debugging

✅ Production-Ready Architecture

• Action-Based Routing: draft_write, draft_rewrite, send, close operations
• Error Handling: Graceful degradation with fallback templates
• Draft Management: Comparison, versioning, and selection utilities
• Timezone Intelligence: Optimal email timing with business hours respect
• Extensible Design: Easy customization and integration

✅ Advanced Intelligence Features

• Personalization Scoring: 0-100 scoring based on customer data usage
• Engagement Analysis: Customer interaction patterns and sentiment detection
• Readiness Assessment: Outreach readiness scoring with recommendations
• Sequence Management: Automatic follow-up progression (1st -> 2nd -> 3rd -> final)
• Respectful Automation: Smart limits and graceful closure handling

🛠️ Trou

bleshooting

Common Issues

1. "No such file or directory: requirements.txt"

Solution: Make sure you're in the fusesell-local directory:

cd fusesell-local
pip install -r requirements.txt

2. "OpenAI API key not provided"

Solution: Ensure your API key is correct and has sufficient credits:

python fusesell.py --openai-api-key "sk-proj-your-actual-key" ...

3. "No data could be collected from any source"

Solution: Ensure the website URL is accessible and valid:

# Test with a known working website
python fusesell.py ... --input-website "https://google.com" --dry-run

4. "Permission denied" errors

Solution: Use user installation or virtual environment:

pip install --user -r requirements.txt
# OR
python -m venv venv
source venv/bin/activate  # Linux/Mac
venv\Scripts\activate     # Windows
pip install -r requirements.txt

Debug Mode

Enable detailed logging to troubleshoot issues:

python fusesell.py \
  --openai-api-key "sk-proj-abc123" \
  --org-id "test" \
  --org-name "Test Company" \
  --input-website "https://example.com" \
  --log-level DEBUG \
  --dry-run

Dry Run Testing

Test the system without making API calls:

python fusesell.py \
  --openai-api-key "test-key" \
  --org-id "test" \
  --org-name "Test Company" \
  --input-website "https://example.com" \
  --dry-run

📚 Complete troubleshooting guide: TROUBLESHOOTING.md

🏆 Production Status

FuseSell Local is 100% complete and production-ready!

✅ All Components Complete:

• CLI Interface: 25+ configuration options with comprehensive validation
• Pipeline Engine: Complete 5-stage orchestration with business logic
• Data Layer: SQLite database with full CRUD operations and event scheduling
• AI Integration: OpenAI GPT-4o-mini with structured response parsing
• Stage Implementations: All 5 stages production-ready (7,400+ lines of code)
• Documentation: Complete user guides, technical docs, and troubleshooting

✅ Stage Implementation Status:

📁 Directory Structure

fusesell-local/
+--- fusesell.py                 # Main CLI entry point
+--- requirements.txt            # Python dependencies
+--- README.md                   # This file
+--- fusesell_local/            # Main package
|   +--- __init__.py
|   +--- pipeline.py            # Pipeline orchestrator
|   +--- stages/                # Pipeline stages
|   |   +--- __init__.py
|   |   +--- base_stage.py      # Base stage interface
|   |   +--- data_acquisition.py
|   |   +--- data_preparation.py
|   |   +--- lead_scoring.py
|   |   +--- initial_outreach.py
|   |   +--- follow_up.py
|   +--- utils/                 # Utilities
|   |   +--- __init__.py
|   |   +--- data_manager.py    # SQLite database manager
|   |   +--- llm_client.py      # OpenAI API client
|   |   +--- validators.py      # Input validation
|   |   +--- logger.py          # Logging configuration
|   +--- config/                # Configuration
|       +--- __init__.py
+--- fusesell_data/             # Local data storage
    +--- config/                # Configuration files
    |   +--- prompts.json       # LLM prompts
    |   +--- scoring_criteria.json
    |   +--- email_templates.json
    +--- drafts/                # Generated email drafts
    +--- logs/                  # Execution logs

🔒 Security & Privacy

• Complete data ownership: All customer data stays on your machine
• API key security: Keys are only used for LLM calls, never stored
• Input validation: Prevents injection attacks and validates all inputs
• Local processing: No external dependencies except for LLM API calls

📚 Additional Documentation

User Documentation

• QUERYING_GUIDE.md - Managing multiple sales processes
• TROUBLESHOOTING.md - Common issues and solutions

Developer Documentation

• DEVELOPER_GUIDE.md - Customization and extension guide
• API_DOCUMENTATION.md - Technical API reference
• TECHNICAL.md - Architecture and technical details
• DATABASE.md - Database schema reference

Reference Documentation

• business_logic.md - Business logic and orchestration rules
• CHANGELOG.md - Version history and updates

🚀 Ready for Production Use

• End-to-End Pipeline: Complete sales automation workflow
• Local Data Ownership: Full privacy and control
• AI-Powered Intelligence: Personalized and context-aware
• Integration Ready: Database events for external app integration
• Comprehensive Testing: Dry-run mode and extensive error handling

💡 Performance Tips

1. Use Dry Run for Testing

Always test with --dry-run first to validate your configuration.

2. Optimize API Usage

• Use appropriate --llm-temperature (0.2-0.8)
• Set reasonable --llm-max-tokens limits
• Consider stopping after specific stages for testing

3. Batch Processing

For multiple leads, use different --execution-id values:

python fusesell.py ... --execution-id "lead_001"
python fusesell.py ... --execution-id "lead_002"

4. Data Directory Management

Use custom data directories for different projects:

python fusesell.py ... --data-dir "./project_a_data"
python fusesell.py ... --data-dir "./project_b_data"

🤝 Support

For issues, questions, or contributions:

• Check the troubleshooting guide for common issues
• Review the technical documentation for advanced usage
• Refer to the business logic documentation for workflow details
• Contact the development team for custom requirements

FuseSell Local - Complete AI Sales Automation, 100% Local, 100% Private 🚀