Project · Hackathon winner
Loomis - AI Design Co-Pilot for Adobe Express
A canvas-aware Adobe Express assistant that recommends relevant images, backgrounds, illustrations, and GIFs with one-click insertion.
Problem
Designers often leave the creative canvas to search for images, GIFs, or visual inspiration. Loomis moves that discovery into Adobe Express so the current design context stays intact.
How it works
Loomis scans the current canvas or an uploaded image, uses Gemini Vision to infer the theme, style, composition, and intent, and then recommends useful images, backgrounds, illustrations, and GIFs. Selected assets can be inserted into Adobe Express in one click.
Lit and Spectrum Web Components provide an Adobe-native interface. The Adobe Express SDK reads and updates the canvas, while Tenor and Unsplash supply visual assets through an orchestration layer that handles compatibility, attribution, categories, loading, and pagination.
Outcome
The team built a working native add-on under a 24-hour hackathon constraint. Loomis won Most Efficient Use of AI for Coding at the Adobe Express Add-ons for Enterprise Hackathon.