Inspiration
While working with Adobe Express, I often found myself struggling to make my designs better. I would frequently take screenshots of my work and consult various AI tools for feedback, which was a tedious and time-consuming process. This back-and-forth between applications made me realize - why not bring the AI feedback directly into Adobe Express? This sparked the idea for DesignMuse: an add-on that could analyze my designs and provide instant feedback right within the Adobe Express workspace, eliminating the need to switch between multiple tools and making the design improvement process much more streamlined.
What it does
DesignMuse is an Adobe Express add-on that serves as your personal design critic. Users can capture their current workspace and ask specific questions about their design, such as "How can I make my text more readable?" or "Is my color scheme professional enough?" The add-on then uses Google's Gemini Pro AI to analyze the design and provide targeted, actionable feedback. Whether you're stuck on typography, struggling with layout, or unsure about color choices, DesignMuse offers specific, context-aware suggestions to improve your work.
How I built it
I built DesignMuse using Adobe Express's Add-on SDK as the foundation, integrating it with Google's Gemini Pro API for AI analysis. The front end was crafted with HTML, CSS, and JavaScript, featuring a clean interface that matches Adobe's design aesthetic. I implemented a document sandbox architecture to securely capture the workspace state and created a custom prompt engineering system to ensure the AI provides relevant, actionable design feedback. The application uses async/await patterns for smooth API interactions and includes comprehensive error handling for a reliable user experience.
Challenges I ran into
The first major challenge was mastering the workspace capture process in Adobe Express to ensure high-quality design captures for AI analysis. Another significant hurdle was integrating Google's Gemini API, as this was my first time working with it. Finally, prompt engineering proved challenging - it took numerous iterations to craft a prompt that would consistently generate specific, actionable design feedback rather than vague suggestions.
Accomplishments that I'm proud of
I'm particularly proud of successfully building my first Adobe Express add-on from scratch, learning to navigate the SDK and understanding the sandbox architecture. Additionally, I'm proud of successfully integrating Google's Gemini AI - while it was challenging to work with a new API, seeing it analyze designs and provide meaningful feedback was incredibly rewarding. The fact that the tool can now seamlessly capture a workspace, process it through AI, and return specific design suggestions all within Adobe Express makes all the learning curves worth it.
What I learned
Through this project, I learned how to use Adobe Express's add-on CLI, set up a local development server for testing, and work with the Document Sandbox API (express-document-sdk). These skills provided me with a solid foundation in Adobe Express add-on development.
What's next for DesignMuse
The future roadmap for DesignMuse includes several exciting features:
- Enabling AI to directly make suggested changes to your workspace
- Implementation of design history tracking to show how suggestions improved the design
- Integration of multiple AI models for comparative feedback
- Adding style transfer capabilities to automatically apply suggestions
- Building a community feature where designers can share their improvement journeys
- Creating customizable feedback templates for different design types (branding, social media, print, etc.)
Log in or sign up for Devpost to join the conversation.