Home — Projects — rabbit r1

about

projects

experiments

Rabbit

2024

Product Design

ROLE

product design intern

TIMELINE

May - August 2024

SKILLS

User research

Interaction Design

Conversation Design

Prototyping

During my time at rabbit Inc., I worked on shaping the kids’ mode experience for the r1 AI device. As the product design intern, I led user research with children and parents, translated insights into interaction patterns and safety logic, and collaborated closely with design, product, and engineering to prototype and test new voice-based experiences.

Details of this work are confidential, but I’m happy to share more in conversation.

Rabbit

2024

Product Design

ROLE

product design intern

TIMELINE

May - August 2024

SKILLS

User research

Interaction Design

Conversation Design

Prototyping

THE PROBLEM

Designing AI For Play Instead of Utility

Kids want a playmate, but parents want safety and control. How do you design one product that's genuinely fun and genuinely trustworthy?

THE PROBLEM

Designing AI For Play Instead of Utility

Kids want a playmate, but parents want safety and control. How do you design one product that's genuinely fun and genuinely trustworthy?

MY ROLE

I was the primary designer leading Kids Mode, reporting directly to the CPO. Starting from a one-line brief with no established roadmap, I scoped the project, ran all user research, designed the interaction framework, led voice engineering collaboration, and validated through three rounds of testing.

Research & Discovery

RESEARCH METHODOLOGY

  • Observed and interviewed a group of children between 6-12 years old

  • Interviewed their parents about safety concerns

  • Reviewed existing kids' companion toys

KEY RESEARCH INSIGHTS

We synthesized our research insights into 3 key points

We synthesized our research insights into 3 key points

1. Imagination beats instruction

Kids preferred interactions where r1 felt like a character or companion, not a traditional assistant. Features framed as imaginative play were consistently more engaging.

2. Short, reactive interactions sustain attention

Children lost interest quickly during long monologues. Fast response time, turn-taking, and interruption mattered more than narrative depth.

3. Safety must be embedded

Parents were concerned about AI safety but did not want kids to feel monitored. They preferred behind-the-scenes safeguards, such as content boundaries and parental dashboards.

Designing the Interaction Framework

DESIGN PRINCIPLE

Interactions are the focus instead of the features themselves.

Interactions are the focus instead of the features themselves.

From early research and testing, one pattern became clear: Kids don’t want features. They want ways to play. Rather than designing isolated features, I proposed organizing kids mode around interaction categories — reusable patterns that could scale safely over time.

INTERACTION FRAMEWORK: 4 WAYS TO PLAY

We defined four major interaction categories, each representing a different mode of engagement with r1.

We defined four major interaction categories, each representing a different mode of engagement with r1.

This framework allowed us to prototype quickly while maintaining consistency, safety, and clarity across features.

It also resolved a core ethical tension: when a child is clearly doing imaginative roleplay in Play Worlds, the AI's character feels like make-believe, not a fake friend. The boundaries between "pretend" and "real" are baked into the structure.

PLAY WORLDS

Stories built together through voice.

Create immersive and imaginative spaces where voice = world-building tool.

Child: “We’re in a forest and it’s dark.”
r1: “I hear leaves crunching. Do you want to move quietly or shout?”

  • Adventure Quests

  • Character Hotline

  • Time Travel & Fantasy Worlds

  • Sound-Only Stories

DISCOVER

PLAY TOGETHER

COMPANION

Prototype & Testing

EARLY VALIDATION

We quickly prototyped experiences across all four categories and tested them with children.

We quickly prototyped experiences across all four categories and tested them with children.

  • rabbit tamagotchi — voice-based companion that encourages care and play.

  • adventure quest — imaginary and interactive adventures shaped by your choices

  • sleepy stories — calm and generated bedtime stories read aloud

  • carrot encyclopedia — voice-first learning through playful curiosity.

Key Learnings

  • Kids stayed engaged longer with reactive dialogue than scripted stories

  • Response speed mattered more than content richness

  • Framing r1 as a peer increased trust and playfulness

  • Voice tone had a significant impact on perceived friendliness

These findings reinforced our focus on interaction design over feature complexity.

Voice Design & Engineering Collaboration

A significant portion of my internship involved collaborating with external voice engineers to develop kid-appropriate voices for r1. I was the sole designer on the Rabbit side leading this partnership.

I found that younger children (ages 6-8) preferred higher-pitched, more energetic voices, while one older child found that same voice "extremely uncomfortable" and preferred something more mature and genuine. We ultimately developed two voice options: one more playful and energetic, another more grounded.

Impact & Reflection

IMPACT

  • validated MVP through 3 rounds of testing, 45% prototype engagement increase

REFLECTION

1. "Kids" is not one audience

I started this project thinking about "kids ages 6-12" as a single group. Research quickly taught me otherwise. A 6-year-old and an 11-year-old have completely different cognitive abilities, social needs, and preferences. The voice testing made this viscerally clear when what delighted one age group made another uncomfortable.

I started this project thinking about "kids ages 6-12" as a single group. Research quickly taught me otherwise. A 6-year-old and an 11-year-old have completely different cognitive abilities, social needs, and preferences. The voice testing made this viscerally clear when what delighted one age group made another uncomfortable.

2. Projects can succeed and still not ship

This was my first experience with a project that tested well but was ultimately cut for strategic reasons. It was frustrating, but it taught me that product decisions happen at multiple levels. Good design isn't enough if it doesn't fit the business direction. I learned to hold my work a bit more loosely.

This was my first experience with a project that tested well but was ultimately cut for strategic reasons. It was frustrating, but it taught me that product decisions happen at multiple levels. Good design isn't enough if it doesn't fit the business direction. I learned to hold my work a bit more loosely.

3. Leading cross-functional work as an intern

This was my first experience with a project that tested well but was ultimately cut for strategic reasons. It was frustrating, but it taught me that product decisions happen at multiple levels. Good design isn't enough if it doesn't fit the business direction. I learned to hold my work a bit more loosely.

This was my first experience with a project that tested well but was ultimately cut for strategic reasons. It was frustrating, but it taught me that product decisions happen at multiple levels. Good design isn't enough if it doesn't fit the business direction. I learned to hold my work a bit more loosely.