
01
Multimodal context routing
Studying how prompts, reference media, project state, and user intent can be routed into cleaner generation flows.
Research
Our research direction is practical: context routing, review loops, workflow memory, and interfaces that help people control AI-assisted output with less friction.

Studying how prompts, reference media, project state, and user intent can be routed into cleaner generation flows.

Designing product patterns for comparing outputs, checking quality, and making AI-assisted work easier to revise.

Exploring how users can preserve direction, decisions, and useful context across a sequence of creative tasks.