Mobile UI Design
YEAR
2024-25
TIMELINE
3 Months
SCOPE
Mobile UI Design
Information Architecture
UX Strategy
Prototyping
TOOLS
Figma
Overview
Recruited by Duke and Stanford founders to design the conceptual iOS MVP for Mindscape, a mental health journaling app powered by AI. I defined the entire UI from scratch: onboarding, journaling, analytics, and provider communication flows.
Context
Most mental health apps are either too surface-level to be useful or too clinical to feel human. Mindscape sits in between as a daily journaling practice where AI learns your patterns, asks better questions over time, and connects those insights to your therapist's dashboard.
The Challenge
Mindscape had no existing design system, no prior UI, and no reference point. And the domain raised the stakes: in a mental health product, a poorly timed prompt or clumsy interface can cause real harm, so every decision carried more weight than usual.
The product also needed to serve two user types at once: a patient on mobile wanting something warm and frictionless, and a therapist on web needing clinical precision across multiple patients. Each required a completely different approach to hierarchy, tone, and interaction, and neither could be designed in isolation.
User Flow
Before any screens were drawn, I mapped the full user journey to establish information architecture and surface the key structural decisions → from where the nav tabs came from to how the journal connects to the user’s insights.
Lo-Fi Wireframes
Lo-fi covered twelve screens across the full MVP. The deliberately flat palette meant every layout decision could be evaluated without visual noise. This is the stage where the structural decisions were made and tested.
From Structure to Solution
The lo-fi stage locked the architecture: what screens exist, what they contain, and how they connect. With that foundation set, each structural decision shaped how the hi-fi was built. The four decisions below each have a direct visual consequence in the final screens.
Design Decisions
Four tabs, not five. And settings, instead of profile.
The nav shows home, journal, insights, and settings. That maps directly to the core loop of writing, reflecting, and gaining insight, with everything secondary tucked into settings. Nothing in the nav competes with the daily habit.
The journal entry carries the AI prompt forward.
Tapping "Answer Prompt" on the home screen takes the user straight into "Today's reflection" with the same prompt at the top, tagged "based on your journals." Mood selection, a 1000-character text area, and tags follow. The flow from home to entry to save is uninterrupted.
The entry archive names the entries, not dates.
Each entry appears as a named card, "Work stress & boundaries," along with a mood badge and tags below. The journal reads as a personal record, not a log.
The confirmation screen makes the streak the moment.
After saving, a flame icon and bold "4 DAY STREAK" card takes centre stage, with reflected themes below and two CTAs: "Done for Today" and "View Insights." The lo-fi used a dot tracker and affirmation. The hi-fi makes the streak the emotional hook because it's what brings users back the next day.
The insights screen turns journaling into self-knowledge.
A 30-day mood trend chart and a ranked theme breakdown, such as anxiety 68%, work stress 45%, gratitude 30%, sleep 22%, show users patterns they couldn't see from individual entries alone. Writing daily builds the data. The insights screen makes it legible.
The Chatbot was cut.
The original brief included a CBT-trained AI chatbot. A language model misreading emotional context in a mental health setting isn't a minor UX error, because it can cause real harm. It was replaced with AI-generated prompts built from the user's own journal history. Safer, and more honest about what the product actually knows.
Dual-User System
The same journal data surfaces differently depending on who is viewing it. Patients see a personal, prompt-driven mobile experience. Therapists see a clinical web dashboard with mood trends, AI flags, tools to assign tasks and track treatment goals.
What's Next
Once usage data confirms writing is the dominant daily action, the journal tab becomes a floating action button. Community scheduling and a local provider map would justify a fifth nav item. Wearable integration would feed passive sleep and activity data into the analytics layer, giving the AI more signal over time.
Interested in hearing more?
I'd love to chat about my process.
Feel free to reach out at sophbaedesign@gmail.com