Sequel Sift

An ETL and ML pipeline for extracting startup data, constructing quantitative tables, and calculating investment risk factors and Sequel Scores.

Quick Start

Prerequisites

• Python 3.6+
• pip package manager
• Internet connection for web scraping

Installation

1. Create and activate a virtual environment

# Create virtual environment
python3 -m venv venv

# Activate virtual environment
# On Unix/macOS:
source venv/bin/activate
# On Windows:
venv\Scripts\activate

2. Install dependencies

pip install -r requirements.txt

3. Launch the data extraction

python sift.py

Workflow

Data Extraction

• Fetches web content from startup company websites
• Handles URL normalization and retry logic for reliable scraping
• Extracts key information from HTML content:
  • Company names
  • Company descriptions
  • Founder information
  • Product details and features
  • About page content

Data Processing

• Text cleaning and normalization
• Natural Language Processing for entity extraction
• Proper noun identification for company names
• Founder name extraction from context
• Product information categorization

Features

• Robust retry mechanism with exponential backoff
• Intelligent company name extraction using NLP
• Multi-source information gathering
• Duplicate removal with order preservation
• Error handling and logging

Output Format

Data is returned in a pandas DataFrame with the following columns:

• domain: Website URL
• company_name: Extracted company name
• description: Company/product description
• founders: Set of founder names
• product_info: Dictionary containing:
  • products: List of product names
  • features: List of product features
  • descriptions: List of product descriptions

Testing

Run the test suite:

python test_sequel_sift.py

Project Structure

sequel_sift/
├── sift.py            # Main implementation
├── test_sequel_sift.py # Test suite
├── requirements.txt    # Dependencies
└── README.md          # Documentation

Dependencies

• requests: Web scraping
• beautifulsoup4: HTML parsing
• nltk: Natural language processing
• pandas: Data organization
• re: Regular expressions
• urllib: URL handling

Next Steps

• Implement OpenAI integration for enhanced information extraction
• Add ML models for risk factor analysis
• Develop Sequel Score calculation
• Create data visualization components
• Build API endpoints for data access

References

• 5 Q's for Philipp Omenitsch CTO of Sequel

Author

Emmanuel Ezenwere

Version

1.0.0

Q&A

What is our goal with this website:

• Efficiently identify promising startups that match Sequel's investment thesis
• Create a data-driven approach to early-stage investing
• Generate insights about successful startup patterns
• Build a community of founders and investors
• Create a tool to help founders improve and optimize their pitchdecks

How would you improve this product:

• Add competitive analysis features
• Create founder success pattern recognition
• Add automated due diligence checklists
• Add more AI coach product pictures to the website
• Extend the capabilities of Sequel's AI coach and Sequel Score Model, Sequel Score Model Proposal.pdf attached.
• Implement robust data science experiment design structure to test effectiveness of Sequel AI coach , A/B Tests.
• Use user feedback to improve the product until Product Market Fit

How can we combine Pitch Decks and Website Crawling:

Data Integration:

Cross-validate claims made in pitch decks with website data Track changes in company positioning over time Verify team information and growth Compare market sizing claims with actual market presence

Analyze Opportunities:

Track execution against pitched roadmap Identify red flags and discrepancies Monitor product evolution Validate customer testimonials Track milestone achievement

Automated Insights:

Generate comprehensive company profiles Create growth trajectory predictions Identify market positioning shifts Monitor competitive landscape changes Track product development velocity