CareYaya
An 80-year-old lost the only person they saw all week. I redesigned the system so caregivers stop quitting.
Role
Product Designer Intern
Timeline
Sep – Dec 2023
Company
CareYaya Health Technologies
Outcome
94% NPS on caregiver training redesign
Imagine you're 80 years old. Your caregiver is the only person you see all week. Then one day they stop coming.
CareYaya matched pre-med students with elderly patients. For the students, it was clinical experience. For the patients, it was companionship. Sometimes the only human contact they had. But students were drowning in coursework and caregiving at the same time. When they burned out and quit, a patient didn't just lose a service. They lost their person.
Stakeholder cards: STUDENTS (pre-med, 1 in 10 are caregivers, juggling school + care, burning out), PATIENTS (elderly, depend on continuity, lose their person when student quits), THE SYSTEM (if students burn out, the entire care network breaks). Make the human cost visible.
The obvious fix: automate the busywork so caregivers have more time.
If students are overwhelmed, give them fewer things to manage. I built a system to automate the repetitive parts of their study and care routines. Schedule optimization. Automatic reminders. Background task management. If we could save them time, they'd have more bandwidth for caregiving.
Screenshot of the automation approach. Show the automated scheduling, the smart reminders, the background task management. It should look like a reasonable, well-designed solution. The recruiter needs to think 'yeah, that makes sense' before the failure hits.
Automation created more stress, not less.
The automated system made decisions students didn't understand. They couldn't see why something was scheduled the way it was. They didn't trust it. Instead of reducing cognitive load, I added a new thing to worry about.
So I tried the opposite. Hide the complexity entirely. Simplify the UI so students wouldn't see the system at all. But hiding information from people who need to make decisions is worse than showing too much. They felt even less in control.
Two-panel failure sequence: ATTEMPT 1: 'Automated the background' with red X. 'Created decisions students couldn't see or trust.' ATTEMPT 2: 'Hid the complexity' with red X. 'Students couldn't verify what was happening.' Show the progression: two reasonable ideas, both wrong. The recruiter should feel the frustration.
The system didn't need to be faster. It needed to be understood.
The real problem wasn't time. It was clarity. Students didn't need fewer decisions. They needed to understand the decisions they were making. When I reframed the settings so they could see the system clearly, everything changed.
Before/after slider: LEFT: The hidden/automated interface where students can't see what's happening. RIGHT: The redesigned interface where the system is transparent. Students see their schedule, understand the logic, and adjust with confidence. Annotate: 'Let users see the system, not just the output.'
Progress tracking that felt like theirs, not the system's.
Incorporated Anki-style graphs that students already knew and loved. Progress tracking became familiar and motivating instead of another thing to learn. I used a pattern they already trusted.
Screenshot of Anki-style progress tracking: heatmap, study streaks, daily stats. Annotate: 'Used a pattern students already trust. Making progress visible in a familiar way turned data into motivation.'
Caregivers stayed 3x longer. Patients kept their person.
Once students understood the system, they could balance school and caregiving sustainably. They didn't burn out. They stayed. And patients kept their caregivers.
92%
Reported better fit with school + care routines
78%
More users completed 3+ day study streaks
67%
Started a session within 10 seconds of opening
8 in 10
Said it helped reduce stress about falling behind
Being wrong twice was the most important part.
A junior designer ships the first solution and moves on. I tried the obvious fix. It failed. I tried a smarter fix. It failed too. The answer wasn't about speed or simplicity. It was about trust. The system needed to be understood before it could be used.
