👻 Please follow me for new updates Dev.to | Github | Github Org
😉 Please join our discord server Discord
💫 Please have an interesting relationship with me Telegram | Email

HOA Document Analysis AI Agent

An intelligent document analysis system that leverages OpenAI's Assistant API to extract, analyze and summarize key information from Homeowners Association (HOA) documents.

Features

• Automated processing of multiple document formats (PDF, Word, TXT, MD)
• Smart information extraction using authority-based ranking system
• Vector store integration for efficient document search
• Comprehensive analysis across 20 key HOA topics
• Detailed answers and concise summaries for each topic
• JSON output for easy integration

Key Capabilities

• Document hierarchy enforcement based on authority ranking
• Intelligent handling of conflicting information
• Citation tracking and source documentation
• Token limit management through iterative processing
• Error handling and retry mechanisms

Configuration

Key settings in main.py:

HOA_DOCS_DIR = "./input/hoa_documents"  # Input documents location
MODEL_NAME = "gpt-4o-mini"              # AI model selection
TEMPERATURE = 0.1                       # Response determinism (lower = more focused)
OUTPUT_DIR = "./output"                 # Results location

Usage

1. Place HOA documents in the input/hoa_documents directory
2. Run the analysis:

python main.py

3. Find results in the output directory:

• {timestamp}-summary.json: Condensed findings by category
• {timestamp}-answers.json: Detailed analysis with sources

Document Authority Ranking

Documents are prioritized in this order (1 = highest authority):

1. CC&R Amendments
2. CC&Rs
3. Bylaws
4. Articles of Incorporation
5. Operating Rules [...]

Analysis Categories

The system analyzes 20 key areas including:

• HOA Name and Official Details
• Financial Information (Dues, Increases, Health)
• Rules and Policies
• Management and Operations
• Community Features and Maintenance
• Legal and Insurance Matters

Requirements

• Python 3.x
• OpenAI API access
• Required packages:
  • openai
  • python-docx
  • PyPDF2 (for PDF processing)

Output Format

The system generates two JSON files:

1. Summary Table:

[
  {
    "Category": "Category Name",
    "Findings": "Concise summary",
    "Source": "Source document path"
  }
]

2. Detailed Answers:

[
  {
    "question": "Original question",
    "answer": "Detailed response",
    "summary": "Brief summary",
    "source": "Source documents",
    "source_ids": ["file_ids"]
  }
]