
Designed Cody AI chat UX used by 50K+ developers. Specializing in AI-native interaction patterns.

Redesigned Cody AI chat UX, tripling daily queries and doubling retention
Cody AI had a basic chat panel that developers used for only 2.3 queries/day on average. The interface felt disconnected from the code editor. 68% of users tried it once and never returned because answers lacked actionable code context.
Redesigned the interaction model around three surfaces: inline ghost text for quick completions, a contextual sidebar that auto-attaches relevant files, and a command palette for complex multi-step tasks. Ran 24 usability sessions with developers across 4 company sizes to validate each surface.
Daily queries per user increased from 2.3 to 7.8. Day-7 retention jumped from 32% to 61%. The inline surface alone drove 45% of all interactions. NPS went from +12 to +47 among active users.

FigJam boards with 10+ simultaneous editors became chaotic. Users reported losing track of who was editing what, and accidental overwrites were a top complaint.
Designed a presence-awareness system with colored cursor trails, activity heat zones, and a "follow mode" that let users shadow a specific collaborator. Prototyped 6 variants and A/B tested with 2K teams.
Accidental overwrites dropped 58%. Session length on collaborative boards increased 23%. Follow mode was adopted by 31% of enterprise teams within the first month.