Project Overview
Carely is a telehealth platform connecting patients with doctors and therapists for on-demand consultations.
I focused on two key moments in the patient journey: finding the right provider and preparing for a consultation. I redesigned both experiences as AI-assisted flows, exploring how conversational AI could reduce manual effort, gather more useful context, and help patients feel more prepared before connecting with a provider.
My Role
I worked as one of three product designers on an NDA'd telehealth app. The AI-assisted discovery and check-in shown in this case study are my own, designed independently on top of it.
Research Plan
How I built a data-driven product
Patients want less friction when finding a provider and preparing for a consultation. Conversational AI can help, but it needs to remain transparent and optional.
Persona & Empathy Map
Using insights from competitive research and published studies, I developed a persona and empathy map to represent the target audience.
Final Deliverables
I designed:
Visual Style
Feature Decisions
Five choices about how the AI shows up, asks, listens, and hands off, made from what the research kept surfacing.
Always tell people it's AI
People trust AI health tools more when the AI is upfront about being AI and explains itself clearly, not just when it happens to be accurate.
Typing is always an option
Not everyone wants to talk out loud about a health problem, so I let people type their answers instead of forcing them to use voice.
Patient approves it before it's sent
Most AI health chatbots don't clearly tell you when you're talking to AI instead of a real doctor, so before anything gets sent to the doctor, the patient has to look it over and approve it first.
Let the patient finish talking
Doctors interrupt patients after about 11 seconds on average, so I built a screen where the patient can fully explain what's wrong without anyone cutting them off.
Confirm the doctor actually read it
Patients often have no idea if their doctor actually read what they sent in before the appointment, so I added a screen that tells the patient "yes, your doctor read this" once they have.
Finding a provider: browse, or get matched
Find Consultant is the standard way in, browse and filter by category. Once a patient describes what's going on, Your AI Matches shows up instead, ranked and explained.
Explaining what's wrong: on the paid clock, or before it
Audio and video consultations bill per minute from the moment they connect, so explaining symptoms out loud eats into paid time. The AI check-in captures all of that first, for free, before the clock starts.
Supporting Screens
A few more screens to round out the picture, the rest of the app is under NDA.
Reflection
