Designing an AI Companion That Feels Emotionally Intelligent

The Problem

Most AI chat products are designed for transactional exchanges: ask a question, receive an answer, and move on. Robyn needed to support vulnerable, ongoing conversations while helping users understand the emotional patterns emerging over time. The experience had to feel warm and personal without making the technology feel intrusive, clinical, or unclear.

The Goal

Create a trustworthy AI companion that encourages users to open up, turns conversations into useful guidance, and provides meaningful reasons to return beyond the chat.

The Scope

As Lead Product Designer, I:

• Defined the product strategy, information architecture, core user flows, and interaction model for a new 0→1 mobile experience.
• Led UX and UI design across onboarding, voice and text conversations, emotional insights, relationship features, and retention experiences.
• Built the design system and partnered with founders, brand designers, and engineers through implementation.

User Tension

Personalization required users to share sensitive information, but asking for too much too early could undermine trust before Robyn had demonstrated its value. The conversation was the product’s primary interaction, but chat alone did not create enough visible, cumulative value to support long-term engagement.

Strategic Pivot

My team and I designed Robyn as more than a chatbot. Conversations became the foundation of a broader emotional intelligence system, feeding into evolving profiles, weekly reflections, trends, and actionable guidance. Sensitive questions were introduced progressively, with contextual explanations showing why the information was needed and how it would improve the experience.

A More Human Conversational Experience

Robyn remains visually present throughout the conversation, creating a stronger sense of continuity than a traditional messaging interface. Voice and text exist within one experience, supported by clear listening, thinking, and responding states that help users understand what the AI is doing.

Insights That Feel Supportive, Not Clinical

I designed an insight system that translates complex emotional and behavioral information into approachable reflections. Plain language, focused visualizations, and actionable recommendations help users recognize patterns without feeling analyzed or diagnosed. Insights include emotional patterns, attachment style, triggers, defense mechanisms, growth opportunities, and emotional strengths.

Value That Extends Beyond the Chat

Features such as the Emotional Profile and You, Unwrapped transform individual conversations into an evolving record of the user’s emotional life. Personalized summaries, trends, reflection prompts, and action plans give users tangible value between conversations and a reason to return.