An App That Removes Its Own Features as Memory Loss Advances
A caregiver sets the memory-loss level, 1 to 5, and the app removes features as it rises
Every care network isolated in its own tenant; patient tablets kiosk-locked
10,000+ engineering hours from 2022 to 2025
About Memory Machine
- Geography
- Australia
- Industry
- HealthTech
- Services
- HealthTech Software Development
Memory Machine is an Australian health-tech startup that builds for people living with memory loss, and for the caregivers and clinicians around them.
SoluteLabs partnered with Memory Machine to build the platform that connects patients, caregivers, and healthcare professionals: a kiosk-locked tablet app for the patient, a companion app for the caregiver, and one backend holding each care network apart from every other.
Solution
We built five pieces: the patient's tablet app, the caregiver's mobile app, a multi-tenant backend that keeps each care network separate, the cloud and delivery pipeline underneath them, and the third-party integrations that carry messaging, device lockdown, and payments. The constraint that shaped all five was the same one the product exists for. The patient's capability changes over time, and the interface has to change with it.
Patient Tablet App
Caregiver Mobile App
Secure Multi-Tenant Backend
Cloud & DevOps
Third-Party Integrations
Deliverables
Accessible Tablet Experience
- A simple kiosk-mode interface that patients can use on their own, with large targets and very little to navigate.
- Medication, appointment and routine reminders reach the patient even when the app is closed.
- GPS-tagged “Find My Things” helps patients locate personal belongings.
Caregiver Companion App
- Mobile dashboards allow caregivers to manage routines, monitor patients, and receive real-time updates and notifications.
- Tracks memory performance and behavioural patterns, so caregivers and clinicians can see change over time.
Secure, Scalable Backend
- Every patient and caregiver network is its own tenant, with encrypted storage and strict access rules, so no caregiver can reach a patient outside their own network.
Reliable Communication Layer
- Chat and video calling between patients and caregivers, stabilised after the vendor integration first put crashes and dropped connections into the app.
Optimized Cloud Infrastructure
- AWS-hosted architecture with ECS, RDS, and CloudWatch for high availability, performance, and cost-effective scalability.
Tech Stack
A standard stack, with two choices specific to this product. Hexnode locks the patient's tablet to a single app, and CometChat carries calls and messaging, which we stabilised by getting fixes from the vendor's own team.
Implementation Highlights
Discovery Sprint
CI/CD setup, infrastructure design, and environment configuration.
Sprint 1-2
Authentication, caregiver onboarding, and initial patient registration.
Sprint 3-4
Stripe integration, user roles, and payment management.
Sprint 5-6
Notification engine, reminders, and in-app communication tools.
Sprint 7-8
Kiosk mode, dashboard analytics, and usability refinements.
Support Phase
Ongoing maintenance, landscape mode support, and feature enhancements.
Outcomes
Four years and 10,000+ engineering hours on a brief with no off-the-shelf answer: an interface that has to get simpler as the person using it loses the ability to use it.
An interface that removes itself
A caregiver set the patient’s memory-loss level, from 1 to 5.
The app withdrew features as the level rose. Most software grows over time, and this one had to shrink on a schedule set by someone else.
Patients kept a working tablet at every stage, never a demoted one.
Isolation by construction
Every patient-and-caregiver network ran as its own tenant in Postgres.
No caregiver could reach a patient outside their own network.
Kiosk mode locked the tablet to the app alone, so it was safe to hand to someone who might not remember what a home button does.
Four years, one partner
Fixed cost, agile, from 2022 to 2025 with a lean team of three to five.
Custom fixes driven out of CometChat’s own team rather than worked around.
Stripe, Bitrise and Vercel carried subscriptions, builds and deploys.
Building for patients whose needs keep changing?
That’s the brief you just read, 10,000 hours of engineering deep. Let’s talk about yours.
