Skip to main content
DocMind is a Python-based document intelligence system built with LangGraph. Follow these steps to get started.

​
Prerequisites

Before installing DocMind, ensure you have:
  • Python 3.13 or higher
  • pip package manager
  • Git (optional, for cloning the repository)

​
Installation Steps

1

Install Dependencies

DocMind uses Poetry for dependency management. The core dependencies are defined in pyproject.toml:
pyproject.toml
[project]
name = "gio-custom-assesment"
version = "0.1.0"
requires-python = ">=3.13,<4.0"
dependencies = [
    "langgraph (>=1.0.7,<2.0.0)",
    "pytest (>=9.0.2,<10.0.0)"
]
Install using pip:
pip install langgraph>=1.0.7 pytest>=9.0.2
2

Verify Installation

Check that LangGraph is correctly installed:
python -c "import langgraph; print(langgraph.__version__)"
Expected output:
1.0.7
3

Set Up Project Structure

DocMind follows this structure:
docmind/
├── components.py       # Core components (Retriever, Judge, etc.)
├── nodes.py           # Workflow nodes
├── workflow.py        # LangGraph workflow definition
├── state_types.py     # State type definitions
├── mock_data.py       # Sample contract data
├── logger.py          # Logging utilities
├── starter.py         # Main entry point
└── test_starter.py    # Test suite
All files should be in the same directory for proper imports.
4

Run Quick Test

Verify your installation by running the starter script:
python starter.py
You should see output like:
================================================================================
Query 1: What are the penalties for late payment?
================================================================================

Response: If payment is not received within thirty (30) days, Client shall 
be assessed a late fee of 1.5% per month (18% annually) on the outstanding 
balance. (See Late Payment Penalties, page 8)

​
Dependency Details

LangGraph is the core orchestration framework for DocMind. It provides:
  • StateGraph: State management across workflow nodes
  • Conditional routing: Retry logic based on judge verdicts
  • Async execution: Efficient async/await patterns
Used in workflow.py:6-34 to build the document intelligence workflow.
Pytest is used for the comprehensive test suite covering:
  • Strategic retrieval tests (Test Set A)
  • LLM-as-Judge evaluation tests (Test Set B)
  • End-to-end workflow tests (Test Set C)
See Testing Guide for details on running the test suite.

​
Troubleshooting

If you encounter ModuleNotFoundError: No module named 'langgraph', ensure you’re using Python 3.13+ and have activated your virtual environment.

​
Common Issues

Import errors with relative imports Ensure all files are in the same directory. The project uses relative imports:
starter.py
from state_types import DocMindState
from workflow import build_graph_workflow
from logger import log_workflow_complete
Python version mismatch DocMind requires Python 3.13 or higher due to modern type hint syntax: 
python --version
# Should show: Python 3.13.x or higher

​
Next Steps

Running Queries

Learn how to run queries and interpret results

Testing

Run the comprehensive test suite

Build docs developers (and LLMs) love

Get started for freeTalk to us