MedVault

Private health records, ready for the next appointment.

Client
Justin and Karen Baker
Year
2026
Our contribution
Product design, React Native development, envelope encryption and AWS Bedrock integration.
MedVault’s home screen with tools to summarise records and prepare for an appointment.MedVault explains what will be shared before the user permits an AI summary.
MedVault on iPhone: personal health records and permission-based AI summaries.

A medical history that stays together

Justin and Karen Baker wanted to make managing health information easier for people living with chronic illness. Records were scattered across appointments, documents and different providers. MedVault brings that information into a personal record the user controls.

We designed and built the app in React Native. It is available on iPhone through the Apple App Store.

Ready for the next appointment

Conditions, medications, allergies, procedures, care teams and test results live alongside the user’s documents and profile. Users can prepare a summary of their records, review it and copy it into a form or take it to an appointment.

The interface keeps navigation clear, text readable and primary actions easy to find. The aim is to make everyday record-keeping manageable even when someone has little energy to spare.

An appointment summary the user can review and copy to the clipboard.A reviewable summary that restates the user’s records, with edit and copy actions.
Appointment preparation: review the summary, edit it, then choose what to share.

All entered data is envelope encrypted

Encryption covers all data a user enters, not just free-text notes. We implemented envelope encryption with password-protected keys: data is encrypted before storage and decrypted on the user’s device.

The database stores the encrypted information. Access to readable records stays tied to the user’s keys, rather than relying on database access controls alone.

AI only with permission

MedVault integrates with AWS Bedrock to access Anthropic’s Claude models for optional summaries. Before a request, the app explains what information will be sent and lets the user inspect it, cancel or explicitly permit processing.

With that permission, the selected information is sent for AI processing. It is not used to train models. The result is returned for the user to review before saving or sharing. This is help organising existing records, not medical advice or diagnosis.

View MedVault on the App Store.

More case studies

Now What

Practical guides and online courses for people navigating separation.

Read more

A Voice-First Connection App for Truck Drivers

A voice-first app that connects truck drivers with each other and puts crisis support one tap away.

Read more

Tell us what you're working on.

Reach out and we will get back to you as soon as we can. We read every message and reply ourselves.

Get in touch