Bushfire Plan Generator

An AI-powered application that helps Australian property owners create comprehensive bushfire survival plans through an interactive assessment process.

Note that these instructions are for a Mac OS machine. You will find equivalent instructions for a Windows machine on the Internet.

Overview

This application uses LangGraph and OpenAI's GPT-4 to guide users through a structured bushfire planning process:

1. Risk Assessment - Evaluates bushfire risk based on location, vegetation, and property characteristics
2. Defense Capability Assessment - Determines ability to defend property during a bushfire
3. Plan Creation - Generates either a "leave early" or "stay and defend" plan
4. Final Documentation - Produces a comprehensive Markdown-formatted bushfire plan

Setup

Once cloned, enter the folder and set up a virtual Python environment:

python3 -m venv venv
source venv/bin/activate

Install dependencies:

pip install -r requirements.txt

Create an OpenAI account and retrieve an API key. Create a .env file in the project root:

OPENAI_API_KEY="<your OpenAI API key>"

Running

To run the application:

python3 main.py

Follow the interactive prompts to:

• Describe why you want to create a bushfire plan
• Answer questions about your property and circumstances
• Make choices about your preferred approach (stay/defend vs. leave early)
• Receive a personalized bushfire survival plan

Features

• Interactive Assessment - Guided questioning process tailored to your responses
• Australian Context - Uses Australian bushfire terminology and safety protocols
• Structured Output - Generates plans using Pydantic models for consistency
• State Persistence - Maintains conversation state throughout the planning process
• Comprehensive Plans - Covers evacuation routes, timing, supplies, and backup procedures

Architecture

Built using:

• LangGraph - Workflow orchestration and state management
• OpenAI GPT-4 - Natural language processing and plan generation
• Pydantic - Structured data validation and parsing
• LangChain - LLM integration and prompt management

Output

The application generates a detailed bushfire plan including:

• Risk assessment summary
• Defense capability evaluation
• Recommended strategy (leave or defend)
• Detailed action steps
• Backup plans for various scenarios

Notes

• Plans are generated based on Australian bushfire safety guidelines
• The application prioritizes human safety over property protection
• All assessments use evidence-based risk evaluation methods
• Generated plans should be reviewed with local fire authorities
• Based on LLM AI models, there is no guarentee that the plan is correct, effective or complete - please review before use
• Once reviewed share with all occupants at the property