Concept Mapper (Prototype)

What

This is a local, AI-inspired (originally built using qwen2.5:7b model for prototype) exploration engine designed to break vague ideas into structured decision paths.

This system simulates a guided thinking process by:

• Generating conceptual categories
• Asking structured questions
• Presenting multiple directions with the options to manually input an option, or reroll the options
• Building decision paths over time

Why

I built this because I wanted to take an abstract concept, and use AI as a tool to break down the concept to the core components, allowing someone to see the structure of their "train of thought". This helps the user think in a straight line, rather than thinking in circles. While most tools jump to straight answers, this focuses on structured exploration of a concept.

How

• A vague concept is provided by the user: "Bake a cake, build a robot, design a car, design a website"
• The system generates a "decision block" (category + question + options)
• The user selects an option, manually injects their own, or rerolls the options (disabled in portfolio version).
• The system updates the state, and continues exploration

Architecture components

• Response parsing system (structured text becomes usable data)
• Decision state tracking (Category stores selected value)
• Iterative decision loop
• Logging system for session tracking

How do I run it? What are the requirements?

python main.py (If that doesn't work, try 'python3 main.py')

• Python 3.x

Standard library modules used: -os -datetime

No external dependencies should be required for this version.

Notes (Important)

• This is a portfolio-safe version
• Full backend prompt logic has been intentionally removed
• Model calls have been replaced with a simulated response
• Screenshots may show the current working version using a local AI model (qwen2.5:7b)

This repository focuses on:

• System design
• Flow control
• Parsing structured outputs

It does not focus on model performance, or final output quality.

What can't it do right now?

• This uses a static response instead of a live model
• Reroll logic is partially disabled in this version
• Output is simplified for demonstration

What's in store for this project?

• This project is part of a larger system that is intended to evolve into: Concept > structured understanding > procedural execution

The execution layer is not included.

Author: CGTechWorks