Muse
A concept interface for a creative AI companion: conversation on one side, a living canvas of generated ideas, moodboards and drafts on the other.
Overview
Most AI assistants are a chat box. Muse explores what happens when the conversation and the output live side by side: you talk on the left, and the right side fills with cards you can pin, remix, group and export.
Challenge
Creative work is non-linear. The interface had to keep every generated idea reachable, make remixing a one-click action, and stay calm despite constant streaming output.
Process
- 01
Mapped a creative session end to end: brief, exploration, selection, refinement, export.
- 02
Designed the canvas as a masonry of idea cards with a small set of verbs: pin, remix, merge, discard.
- 03
Built streaming responses with token-level animation so the assistant feels present without being noisy.
- 04
Tested the layout with real briefs to find the right balance between chat width and canvas space.
Solution
A React application with a two-pane layout. The conversation streams from a language model; each response can spawn cards on the canvas, and cards can be sent back into the conversation as context. Motion is handled by Framer Motion for layout changes and GSAP for the streaming text.
Features
Full-Stack E-Commerce Platform
A complete online store with product management, authentication, cart and checkout, orders, database integration and an admin dashboard.
Open case study