Mindee — Emotional Patterns, Visualized
Lead UI/UX · Product · Real-world Product
Lead UI/UX · Product · Real-world Product
Mindee is an AI companion for emotional awareness and regulation.
This is not a concept project — it is a real, App Store–released product shaped by research, engineering constraints, and continuous updates.
I led the product direction, UX research, interaction models, and the full design system across iPad and iPhone.
BACKGROUNDDuring our earliest interviews, a user said:
“I can talk, but I don’t really know what I’m feeling.”
This moment became the foundation of Mindee.
Rather than building another conversational tool, we chose to create a structure for emotions — something that can be seen, held, and returned to.
Each marble represents a feeling with its own texture, weight, and tone — a visual unit the AI can interpret and the user can understand intuitively.
This structure guides the entire product.
I worked as both Lead UI/UX Designer and Product Manager, defining:
-
Product vision, scope, and release roadmap
-
Cross-market research (US & China)
-
Information architecture and feature prioritization
-
The Emotion Curve, CBT Diary, and interaction patterns
-
AI behavior in collaboration with the CTO
- Engineering alignment and handoff (MVP → V1 → continuous updates)
Building for Reality
As the product grew, many ideas were rewritten or reshaped:
- Interaction models adjusted for feasibility
- Emotional inference tuned through testing
- Rendering optimized for mobile performance
- Dual-device rhythm refined (iPad vs iPhone)
- Privacy constraints influencing design decisions
Mindee exists the way it does because it had to work.
Core Interaction
Mindee’s experience is built across three layers:
1 · Awareness
Emotion labeling · First-pass marbles · Gentle AI prompts
2 · Exploration
Emotion Curve · Fragments · Time-based reflection
3 · Regulation
AI suggestions · Breathing modes · CBT-inspired guidance
Users move through these layers naturally as their emotional state shifts.
Mindee’s experience is built across three layers:
1 · Awareness
Emotion labeling · First-pass marbles · Gentle AI prompts
2 · Exploration
Emotion Curve · Fragments · Time-based reflection
3 · Regulation
AI suggestions · Breathing modes · CBT-inspired guidance
Users move through these layers naturally as their emotional state shifts.
Onboarding the Mindee Experience
Onboarding introduces the rhythm of Mindee — slow, calm, supportive.
It establishes the tone before users enter their daily emotional space.
Design Language
Mindee’s visuals avoid both the clinical and the playful.
The lofi mode reflects a quieter emotional space — neutral, soft, uncolored —
a place for users to breathe without pressure.
The lofi mode reflects a quieter emotional space — neutral, soft, uncolored —
a place for users to breathe without pressure.
Extending the System
Features like MBTI personality insights expand Mindee’s emotional framework, offering additional ways for users to understand patterns and tendencies.
Outcomes Based on internal pilots and real-world use:
Features like MBTI personality insights expand Mindee’s emotional framework, offering additional ways for users to understand patterns and tendencies.
Outcomes Based on internal pilots and real-world use:
- 2.3× increase in journaling completion
-
37% improvement in AI emotion-label accuracy
-
30+ high-fidelity screens shipped to engineering
- App Store launch with active version updates
Emotions change quietly.
Products that hold them must learn quietly, too.
Mindee is still learning — with every update, every user, and every moment someone pauses to reflect.
Products that hold them must learn quietly, too.
Mindee is still learning — with every update, every user, and every moment someone pauses to reflect.