Feedstack

A Multimodal UX Feedback System for Web & App Designs

Feedstack is a cutting-edge feedback platform that leverages GPT-4 Vision to deliver intelligent, multimodal UX feedback on web and mobile design mockups. Our system helps designers, developers, and researchers streamline UI/UX evaluation by analyzing visual designs and providing categorized, theme-based recommendations in real time.

Key Features

Chat-based AI Assistant: Interact with a GPT-4V-powered assistant that provides personalized design feedback.
Multimodal Analysis: Upload images of your designs (mobile/web) for analysis.
Design Theme Categorization: Feedback is automatically grouped into key UX themes like Accessibility, Color Contrast, Usability, and more.
Interactive Accordion UI: Feedback is neatly displayed in expandable accordions for better navigation and focus.
Keyword Highlighting (Upcoming): Identify theme-relevant keywords directly in the chat interface with dynamic color-coded highlights.

How It Works

Log in with a participant ID.
Upload your design mockup (PNG, JPG).
Chat with the assistant and ask for feedback.
Explore feedback summaries grouped under design themes.

Use Cases

UX Research and Evaluation
Rapid Design Feedback during Prototyping
Accessibility and Visual Hierarchy Testing
Educational Tool for UI/UX Learning

Built With

Frontend: React, HTML, CSS
Backend: Django, Django REST Framework
AI Model: GPT-4 Vision (GPT-4V) via OpenAI API
Other Tools: Python, JavaScript

Project Status

Feedstack is currently in its development phase. New features in development include keyword-theme linking, accordion navigation, and improved design understanding through embeddings.

Documentation: Feedstack Project Site