Designing Emotionally-Aware AI for Memory Sharing

When I joined Shine, its photo AI was powerful—but the experience lacked clarity and warmth. My redesign reimagined Shine Photos as an emotionally intelligent platform for private, AI-powered memory sharing. Later, this evolved into Aura, a photo-based assistant that used LLMs to deliver emotionally resonant insights—deepening trust and connection through AI.

⏳ Timeline

2024–2025

Timeline: 10+ months (2024–2025) across multiple phases
Collaborated with Product, Engineering, and the CEO

👩🏻‍💻 My Role

Lead Product Designer

• UX Strategy, Interaction Design, Visual Design, Prototyping

• Drove the full redesign for Shine v3 (live) and future exploration for v4

• Collaborated closely with Product, Engineering, and the CEO

• Led design systems, onboarding, AI Suggest flows, tagging, and real-time sharing features

• Prototyped new experiences including:

  • Shared Camera (POV-style group capture)

  • Widgets + App Clips for seamless, download-free access

  • Progressive AI prompts for trust and explainability

🔍 The Challenge

Strong AI engine. Weak trust, emotion, and usability.

Users didn’t understand how Shine’s AI worked or what they were supposed to do. Sharing felt clunky. The product was built for utility, but our users wanted something human.

Key Issues:

• AI felt like a black box—no context, no control

• Streams, suggestions, and tagging were confusing

• Emotional value of photos was missing

• Sharing workflows felt like task management

We also faced internal constraints:

• Shifting leadership direction required repeated reframing of priorities

• A full redesign (v4) was paused pre-launch due to resourcing changes

Utility-first layouts, dense text, and a lack of visual warmth made Shine’s V1 core flows feel more like task management than memory sharing. These early screens revealed critical UX gaps in trust, clarity, and emotional engagement—becoming the foundation for a more intuitive, human-centered redesign.

🧠 Research & Personas

Designing for real needs, not just functionality

We interviewed users across group events and found consistent pain points with photo sharing. People wanted help from AI—but only if it felt intuitive, transparent, and emotionally aware.

Olivia, 42 – Family Archivist	Raj, 35 – Group Trip Organizer	Maya, 26 – The Social Snapper
Device: iPhone | Tech Comfort: Medium Needs: Collect and share family event photos across generations Frustrations: Group chats are messy, iCloud is confusing, not everyone is on the same platform What Shine Solves: AI-suggested photos, full-res sharing across devices, Streams that organize memories by event—not thread 💬 “I just want the good photos from the birthday party—without having to ask five people to send them.”	Device: Android + MacBook | Tech Comfort: High Needs: Easily gather and share trip/wedding/event photos Frustrations: Shared albums get ignored, photos pile up, no one wants to organize them What Shine Solves: Clusters duplicates, tags friends automatically, and makes it easy to share selectively 💬 “I always end up being the one chasing down everyone’s photos. I’d rather the app just figured out the best ones to share.”	Device: iPhone | Tech Comfort: High Needs: Instantly share and receive photos from nights out or weekend hangs Frustrations: Messaging apps compress image quality; shared folders feel like extra work What Shine Solves: Shared Camera lets everyone contribute in real time; Suggestions surface the best photos 💬 “I just want last night’s best pics to show up—without chasing the group chat.”

Across all users, one thing was clear: people were open to AI—if it felt intuitive, transparent, and emotionally aware.

🧭 Competitive Research

A quick landscape scan revealed that existing tools were fragmented:

✨ Design Goals

Make Shine emotionally intelligent—not just functional

• Shift from cold utility to warm, human connection

• Clarify how AI works while giving users control

• Introduce small moments of delight

• Align flows with how people actually experience memories

🧱 Design System

Scalable visual + interaction patterns

To support this shift, I built a new design system with:

• A tone-aligned UI kit for clarity and warmth

• Emotion-aware interaction patterns

• Foundations that scaled into Shine v3 and future AI prototypes

🚀 Onboarding Redesign

Build trust from the very first tap

Instead of pushing permissions right away, we started with:

• Friendly, human intros using real photos and names

• Clear previews of photo value before asking for access

• Nudges like “Say hi” to ease social entry

This improved early engagement and positioned AI as helpful, not invasive.

Showcase the onboarding flow — welcoming screens, previews of value, friendly permissions.

🔄 Core Flow

Tag → Suggest → Share

A simplified loop that made sharing feel intentional and seamless.

1. Tag people in photos

2. AI suggests the best moments

3. Confirm + Share in one seamless motion

This reinforced user control and increased confidence in AI.

A simplified loop that made photo sharing seamless. Tagging became intentional sharing. AI suggestions were contextual. One-tap confirmation closed the loop.

Why it matters:

Tagging someone’s face in Shine isn’t just for organization—it’s an intentional act of sharing. By confirming names, users are prompted to share directly with tagged friends, streamlining a process that’s often fragmented. This flow reduces steps, clarifies what happens next, and reinforces trust by showing users exactly who will see which photos.

We also introduced Tagging Achievements—milestone badges that added a playful, social layer to Shine and encouraged participation without pressure.

3-step interaction showing the loop from tagging a face to AI suggesting photos to tapping share.

📸 Stream-Based Model

From static albums to living memories

We introduced Streams—ongoing, dynamic collections based on faces, events, or places. Unlike traditional albums, these evolved as new photos were captured, mirroring how real memories grow over time.

Why it’s different:

Streams replaced static folders with evolving, contextual threads. Users could create them manually or rely on AI-powered Suggestions, which grouped photos by date and location. Empty streams encouraged contribution, while active ones displayed shared moments in rich, ongoing context.

Whether retroactive or real-time, Streams became the social backbone of Shine’s sharing model—designed to feel more like a conversation than a folder.

Animation of a stream growing over time, new photos added with contextual cues

💬 Social Nudges

Subtle touches to humanize AI

To reduce friction and awkwardness, we added lightweight social features that made sharing feel personal—without overwhelming users.

• “Say Hi” moments when friends joined

• Pre-filled DMs and photo suggestions to spark interaction

• Tagging stats and private invites to recognize shared moments

These social touches helped Shine go beyond utility.

“Say Hi” welcome cards made joining feel warm and interactive.

“Photos Shared with You” showed who tagged you and what memories you were in.

Together, these micro-moments fostered a sense of community—turning Shine into a space where sharing felt seen, celebrated, and human.

Tap “Say Hi” → send pre-filled message → friend sees shared photos

🎥 App Clip + Shared Camera

Instant access. Live collaboration. Vision for what’s next.

To reduce friction and support spontaneous sharing, I designed a lightweight App Clip flow that let friends join a Stream, take photos, and contribute in real time—no full download required.

What we shipped:

• Join via link or QR: Opens directly in App Clip

• Quick setup: Just name and number to join

• Smart permissions: Prompted only when needed

• Auto-contribute: Photos instantly added to the group Stream

• Designed for spontaneity: Perfect for events, parties, or casual hangs

This experience became a key differentiator—lowering barriers to entry while preserving trust and privacy.

Enter App Clip → join stream → contribute photos live

Where we were headed (Shine v4 vision):

We expanded the concept into a more immersive, collaborative camera system that supported real-time storytelling:

• Shared Camera (POV-style): Friends capture and contribute photos live into a shared stream

• Shared Cams Hub: Grouped by context—Trips, Parties, Family Moments

• Auto Highlights: AI curates standout shots into ready-to-share stories

• One-Tap Shares: Smart prompts to share moments with the right people

• People & Places Filters: Quickly browse memories by face, trip, or location

• Homescreen Widgets: Launch directly into Streams or highlight favorite moments

• Streamlined Onboarding: Contextual, skip-light, QR-friendly flow

Though Shine v4 didn’t ship, these explorations shaped the long-term vision of Shine as a cross-device, emotionally aware memory system powered by real-time collaboration and lightweight AI cues.

📊 Impact & Signals 

Redesigning Shine Photos led to real engagement and trust gains—both in metrics and user sentiment.

🌟 Key Outcomes

• +48% tag-to-share completion

  The simplified flow helped users move seamlessly from face recognition to intentional sharing.

• -35% onboarding drop-off

  Friendly, value-first entry points—especially via App Clip and invite links—kept more users engaged.

• Higher engagement with AI-suggested streams

  Clearer suggestions and trust cues reduced drop-off and increased stream activation.

• Shared Camera + App Clip approved for roadmap

  These features weren’t fully launched yet, but our prototypes shaped Shine’s direction for real-time, collaborative sharing.

⭐ User Feedback Highlights

Shine v3 averaged a 4.5-star rating across nearly 200 reviews, with recurring praise for:

• AI-powered tagging and clustering

• Streamlined photo sharing

• Automatic album creation

• Friendly, intuitive onboarding

“Love how Shine just gets the good photos. No chasing group chats anymore.”

“Finally—an app that makes sharing feel easy and private.”

🗓️ Reflection

Designing Shine challenged me to balance emotional resonance with AI complexity.

While the backend was intelligent, the real work was making that intelligence feel human—earned through clear flows, gentle prompts, and visible value.

This project taught me to:

• Build trust in AI through micro-feedback and transparency

• Design flows that honor user intent without introducing friction

• Advocate for emotionally aware features—even when roadmaps shift

Not everything shipped, but many of those concepts seeded future thinking. It reinforced a truth I carry forward:

Thoughtful design doesn’t just ship features—it lays the foundation for what’s possible next.

🤖 What Came Next: Aura

From memories to meaning.

The emotional foundations of Shine Photos inspired something more ambitious—Aura, a photo-based AI assistant that reflects who you are through identity insights and gentle nudges.

Built on the same trust-first design principles, Aura used LLMs to surface patterns in how people spend time, connect with others, and capture what matters most.

👉 Read the full Aura case study →