A practical workflow for structured AI collaboration in software development.
This repository documents a workflow where AI is used not only for code generation, but also as a design discussion partner, reviewer, and documentation assistant.
The goal of this approach is to keep human control over system design while using AI to accelerate discussion, implementation, and iteration.
Instead of using AI only as a coding tool, this workflow separates AI collaboration into two roles.
Human (Planner)
Design AI
→ design discussion
→ system structure review
→ documentation organization
Implementation AI
→ implementation planning
→ code generation
Development proceeds through the following cycle.
Idea
→ Design Discussion
→ Documentation
→ Implementation
→ Review
→ Documentation Update
This structure allows design and implementation to evolve together while keeping the system architecture stable.
This repository contains three documents that explain the workflow.
Defines the key terms used in the documents to maintain consistent language across design, documentation, and implementation.
docs/Eng/terminology.md
docs/Kor/terminology.md
Explains the overall development methodology.
Topics include:
- design-driven development
- document structure
- role separation between AI systems
- iterative development workflow
docs/Eng/ai_collaboration_workflow.md
docs/Kor/ai_collaboration_workflow.md
A real example showing how the workflow is applied in practice.
This case demonstrates how:
- a design gap was discovered during implementation preparation
- the system structure was clarified
- a Spec document was created before implementation
docs/Eng/ai_collaboration_case_study.md
docs/Kor/ai_collaboration_case_study.md
Many AI-assisted development workflows focus mainly on code generation.
However, large projects often fail not because of coding problems, but because of unstable design and unclear system structure.
This workflow focuses on:
- keeping design decisions under human control
- using AI to improve design discussion
- organizing knowledge through structured documentation
- ensuring implementation follows clear specifications
This workflow can be applied to many types of projects, including:
- game development
- software systems
- tool development
- AI-assisted solo projects
- small team development
It is particularly useful when:
- AI is heavily involved in development
- the system structure is complex
- the project is expected to grow over time
Hoon Heo
This work is licensed under the Creative Commons Attribution 4.0 International (CC BY 4.0) license.
You are free to:
- share
- adapt
- build upon the material
for any purpose, including commercial use.
The only requirement is proper attribution to the original author.
If you reference or reuse this material, please include attribution such as:
Design-Driven AI Development
by Hoon Heo
or a link to this repository.
This repository shares a workflow developed through real AI-assisted development practice.
The goal is to help others explore more structured and effective ways to collaborate with AI in development.
⭐ If this guide is useful, consider starring the repository.