AI-Powered Experience Design

Leah Wellness

Humanizing the search for mental healthcare by replacing clinical hurdles with compassionate, conversational AI guidance.

Role

Lead UX Designer

Timeline

6 Months

Team

1 PM, 2 Engineers, 1 Researcher

Tools

Figma, Jira

Project Overview

Leah Wellness is an AI-driven startup aiming to reduce the friction in finding the right therapist. I joined to redesign the core onboarding experience, moving from a clinical, form-heavy process to a compassionate, conversational journey.

The Challenge

Seeking help is hard. Patients often don't know clinical terms, and finding an available, in-network provider feels like finding a needle in a haystack. This high friction leads to 'Search Fatigue' — where users give up before ever booking an appointment.

Key Insight

60% of users abandon the search before finding a provider

What Users Told Us

Search Fatigue

It takes a while to find the right therapist... most of it is trial and error.

— Serena

Key Insight

Users are exhausted by the cycle of trying and failing to find a match.

Cognitive Overload

I have to dig through my email and it's all up in my brain somewhere.

— Elena

Key Insight

Fragmented data across portals makes care coordination a mental burden.

The Trust Gap

I'm suspicious of those third party health providers. Who has access to this?

— Kira

Key Insight

Skepticism is the biggest barrier to adopting AI in healthcare.

Core Design Principles

AI as Translator

Translating everyday feelings and symptoms into clinical terminology that matches the right specialist.

Progressive Disclosure

Hiding complexity until needed. Users only see what's relevant to their current step in the journey.

Empathetic Intelligence

Building trust through warm, conversational interactions that feel human and supportive.

Deep Dive 01

From Interrogation to Conversation

The Problem

The original intake was a daunting 20+ input static form. It felt clinical, overwhelming, and triggered anxiety — the opposite of what mental health users need.

"It's hard to just list out all my trauma in a form. I need someone to listen."

— Sommer

The Solution

I replaced the form with a gentle chat interface. This creates psychological safety and mimics the experience of an actual therapy session.

—One question at a time to reduce cognitive load
—Natural language input instead of dropdowns
—Contextual follow-ups based on user responses

Click to flip card

Deep Dive 02

Visualizing Trust in AI Recommendations

The Problem

Users were skeptical of AI-generated recommendations. They asked: "Why did the algorithm pick this person for me?" The lack of transparency eroded trust.

"I don't believe in AI care plans because I don't know if it's accurate or private."

— Serena

The Solution

I designed a two-sided card system that balances quick decision-making with deep reassurance.

Front Side

Match Score, Cost, Next Availability — for quick scanning

Back Side

"Why this match" — explains the AI's reasoning in plain language

Visual Identity

The visual language needed to feel calm and trustworthy. I used soft gradients, generous whitespace, and rounded forms to create a sense of "breathing room" — essential for a mental health product.

Design System — Colors, Typography, Components

Outcome & Impact

<10

Minutes

Reduced clinical intake time from 25+ minutes to under 10.

↓40%

Drop-off Rate

Projected reduction in user abandonment during onboarding.

↑2.3x

Booking Conversion

Improvement in booking conversion based on usability testing.

Key Learnings

Trust is earned progressively. Users need to understand AI reasoning before they'll trust its recommendations.

Conversational UI reduces anxiety. The chat format created psychological safety for sensitive health topics.

Next Project

Yoyo — Hyperlocal Marketplace