AI Design Agent is an autonomous design assistant built on JSON-native workflows. It transforms natural language prompts into structured visual generation instructions and delivers high precision and controllability using Bria FIBO’s advanced parameters for camera angle, lighting, color palette, and style. The system continuously improves its outputs through agentic refinement loops that analyze feedback, adjust parameters, and rerun iterations when needed.
The solution includes a React frontend with a chat interface and image gallery, while the backend features an API layer for request handling, workflow orchestration, and iteration management. Its business logic uses an intent parser powered by a chain-of-responsibility pattern, a strategy-based JSON generator, and an agentic refinement engine with feedback analysis and iteration tracking. It integrates seamlessly with Claude for language intelligence, Bria FIBO for image generation, and a session database for user context.
The AI Design Agent stands out through its full JSON-native generation pipeline, deep controllability, and autonomous agentic workflow—three pillars aligned with the strongest hackathon judging categories. It is designed for product showcases, marketing creatives, social content, and any workflow that benefits from rapid, automated design iteration.
Log in or sign up for Devpost to join the conversation.