Reverie Shell
Rethinking the operating system for people who live in 40 windows at once.
TYPE
Software · Operating System
YEAR
2025
TAGS
Desktop · Design System · Motion
ROLE
Principal Designer
TIMELINE
Sep 2024 – ongoing
CLIENT
Reverie Shell
TEAM
3 designers, 11 engineers, 2 researchers
Overview
Reverie is building a desktop OS for people whose work spans dozens of apps and contexts every hour. The challenge wasn't inventing new metaphors — it was making old ones (windows, files, focus) hold up under modern weight. Every interaction had to feel inevitable, never clever.
Process
Started with a six-week study of how 30 power users actually use their machines. Found that "switching context" cost them 23 minutes per hour on average. Designed a spatial workspace model where related windows cluster automatically. Prototyped 4 motion languages before settling on one that signals state without demanding attention of the users.
Outcome
Closed alpha launched to 200 users in March. Average session length is 4.2 hours — three times the team's target. The motion system has been open-sourced and adopted by two other early-stage OS projects. Public beta is planned for Q3.
Key Decisions
Kept the window as the core metaphor. Users already think in windows — reinventing that would have taxed the exact people we were trying to help. Built focus as a system state, not a mode. A separate "focus mode" would have been one more thing to toggle; instead the whole OS quiets around what you're doing. Open-sourced the motion layer early. Consistent motion was doing the heavy lifting for spatial memory, so we made it a shared primitive rather than a hidden detail.
Conclusion
Reverie taught me that the hardest interface work is often subtraction, not invention. By trusting metaphors people already carried and letting motion and restraint do the explaining, the OS could hold enormous complexity without ever feeling heavy. The alpha numbers matter, but the quieter win is that people forget they're using it at all.
Gallery
