CareCircle: one app for everyday tasks, built on trust
Groceries, rides, medication, bills and family support in one consistent app. Designed for older adults who told me trust matters more than features.
TIMELINE
Aug 2025 to May 2026
ROLE
UX designer and researcher
TYPE
Self-initiated, extended in two courses
RESEARCH
15 older adults interviewed
ON THIS PAGE
Overview
At a glance
Problem
Key decisions
Process
Solution
Learnings
Overview
It started at home. My mother got stuck between apps, payment steps and confusing text confirmations when she ordered groceries. My father got lost in buttons and unclear terms when he booked a ride. Interviews with 15 older adults showed the same pattern: the barrier was trust and fear of costly mistakes, more than ability. So I treated it as a systems problem. CareCircle puts one accessibility-first interface over partner services (groceries, rides, medication, banking), runs every task through the same review, confirm and undo flow, and gives family access only by permission.
What the research showed
15
Older adults interviewed about the apps they use today.
Every one
found managing several apps hard. Most would try a new one if family recommended it.
Trust
was the main worry: data security and trusting a new app, more than whether they could use it.
Why everyday apps are hard
Not a skills problem. A design problem.
Every service has its own app, and every app assumes you are already confident with phones.
Apps for groceries, rides, health and banking each optimise for speed: dense screens, many decision steps, costs that show up late, and accessibility settings that are optional or inconsistent. Each has its own login and its own rules. For someone on a fixed income, one wrong tap can cost real money. The UN expects 2.1 billion people aged 60 and over by 2050, so the gap will only grow.
Three technical decisions, and what each one cost
01
Orchestrate partners, don't own services
Decision: one interface over partner APIs, with every order, ride, refill and bill passing through the same review, confirm and undo steps. Rejected: building each service (cost, regulation) and linking out to partner apps (separate logins and layouts, the exact problem). Trade-off: dependence on third-party APIs and more backend work. Mitigation: each partner sits behind one internal contract, so a failure shows one plain message instead of a broken screen.
02
Undo by design: pay late, not early
Decision: a review screen (who, how much, from where) before any money moves, then a short undo window before the payment settles. Rejected: one-tap checkout, which is faster but unforgiving. Trade-off: payments settle later and the backend must hold pending actions. Why it matters: interviewees feared costly mistakes, so a safe way back makes trying something new less risky. To measure: undo rate and backed-out reviews.
03
Least access first, family by permission
Decision: a linked bank is view-only, the password is entered on the bank's own window, and family roles see only what the user switches on (alerts, bill reminders, balance), with money movement off for family. Rejected: shared logins (no audit, all-or-nothing). Trade-off: more settings to design; I kept it to four plain-language switches. Risk: needs compliance review and a real banking partner to validate.
How I got there
01
Research
Interviewed 15 older adults about the apps they use. Read the research on designing for aging (Pak, Rogers, Czaja, Charness) and Don Norman's principles.
02
Define
Named the real barrier: fear of mistakes and trust, more than ability. Mapped who is involved: the older adult, family caregivers, service providers and health partners.
03
Explore
Compared Amazon, Uber, Apple, Google and bank and health apps. Each is strong alone; none joins daily tasks with low effort. I chose to orchestrate partner services, not own them.
04
Design
Sketched, then wireframed onboarding, then built high-fidelity screens: large touch targets, plain language, voice always optional, the same patterns in every service.
05
Test
Tested the concept in interviews and feedback sessions. Not yet tested with people who have severe vision or hearing loss. That is the first gap to close.
06
Plan
Mapped 24 months: a secure MVP for about 1,000 users, then partner onboarding, family sharing, and scale. My estimate for phase one is $250,000 to $400,000.
One home for daily tasks
One calm home, four clear doors
One interface over many services, with the same review, confirm and undo flow in each.
The home screen greets you by name and offers Groceries, Rides, Health and Circle as large tiles, with My Money just below. Health shows today's medication with a clear yes-or-no confirmation. My Money links a bank as view-only, so the app can show a balance and upcoming bills but cannot move money on its own. Every payment is reviewed first and can be undone. A family member can be given alerts only, and you choose what they see. Watch it run below.
Home
Four large choices and one suggested next step, such as a dose due or groceries running low.
Health
Plain check-ins like "Have you taken your Lisinopril today?" If a dose is missed, family can be told.
Circle
Shop your usual items, see the total first, and keep family close, with sharing you control.
Help
A real person any hour, plus voice tips. Not a chatbot menu.
How it works under the hood
01
Interface
Mobile-first and accessibility-first: large touch targets, high-contrast mode, speech-to-text search, voice always optional, guided step-by-step flows, and onboarding that can be repeated. One component set serves all four services.
02
Workflow
Every task runs the same short flow: choose, review, confirm, undo window, receipt. Groceries, rides, refills and bills share it, so the rules a user learns once apply everywhere.
03
Partner APIs
Grocers, pharmacies, ride vendors, payment systems and health providers connect through secure APIs behind one internal contract. A partner can change, or fail, without the user's screens changing.
04
Data
Holds preferences, service history, reminders, caregiver access settings and transaction records. Health and financial data are kept apart and shared only through explicit permissions.
05
Security and roles
Role-based access for the older adult, family, service providers and health providers. Health data follows HIPAA, financial data follows financial security rules, and personal data follows GDPR where it applies.
06
Cloud and monitoring
Scalable hosting with backups and uptime monitoring, push notifications for alerts, and analytics on service use and accessibility feedback. First target: a secure MVP for about 1,000 users.
The words do the trust work
01
Ask, don't assume
Not a lone "Mark complete" button. Instead: "Have you taken your Lisinopril (blood pressure medicine) today?" Yes or no. If no, we ask why: "I took it but forgot to mark it", "I'll take it soon", "I don't feel well", "I need help." The answer helps the user, and shows family a pattern, not just a miss.
02
Name the cost and the way out
Not "Confirm transaction?" Instead: "Review before you pay. To Electric Co. From Checking ending 4821. $64.20. No extra fees. You can undo this." Who, how much, from where, and a clear exit, all before the button.
03
Say what we can't do
When linking a bank: "View only to start. We can't move money. We never see your bank password. Disconnect any time." Stating limits first answers the worry interviewees raised before they have to ask.
How I would measure success
Tasks
Can a first-time user order groceries or pay a bill unaided? Measured as unaided completion and time to first task in moderated tests.
Undo
How often people reverse a payment or back out of a review step. A falling rate shows the confirmations are understood, not just present.
Trust
Share of users who link a bank, invite family, and keep using the app, alongside support calls per active user. My plan assumes 70 to 85 percent retention; that is a target to test.
What I learned
Simple is not simplified
I first thought the answer was bigger and simpler. The interviews showed older adults want simplicity without being talked down to, help without dependence, and confidence, not just ease.
Start narrow, then widen
Four services at once is a lot of backend, compliance and partner work. Next I would pilot one, medication, with one health system and people who have chronic conditions. Then test with severe vision and hearing loss, and design what users see when a partner API is down. Business projections in my plan, such as 30 to 50 percent lower acquisition cost through partners, are assumptions to test.
Explore more
Hitachi DRE: an AI pipeline dashboard
The Cold Mode
Want the full story?
The screens on this page are my own CareCircle designs, and the research comes from my interviews and course reports. I'd be glad to walk you through the trade-offs, what I would test next, and what I would do differently.
itsmehersunitha@gmail.com
◕‿◕
CareCircle. All your daily needs, in one simple place.