Inspiration

We were inspired by a close friend of one of our teammates, who had been living with Alzheimer’s for over ten years. On the very first day of this project, we planned to speak with her to learn more about her journey. However, we got the sad news that she passed away that same day when the hackathon started. That loss made the project feel deeply personal and sharpened our vision. We weren’t just building a cool MR demo, we were asking how technology might honour memories, support dignity, and ease a very real kind of pain, as suffering from Alzheimer's can bring.

We drew on research from Reminiscence Therapy (RT) and Cognitive Stimulation Therapy (CST), which shows that revisiting meaningful memories and gently exercising attention, language, and problem-solving can help maintain cognition and emotional well-being for people with Alzheimer’s. This combination of personal connection and resseach evidence inspired us to design Memory Garden.

What is the experience?

Memory Garden is a mixed reality, AI-enhanced experience for people living with Alzheimer’s, dementia and memory loss can gently re-experience their life memories through sensory, playful interaction. Designed for the Meta Ray Ban AI Display Glasses.

The user enters a gentle garden of 3D objects generated from their family photos. When they reach out and select an object:

  • The object moves closer and reveals the original 2D photograph linked to that memory.
  • They hear the voices of loved ones telling the story behind that memory.
  • An AI agent can respond with gentle reminiscence and simply keep the conversation going based on the family members' story connected to the. photograph.

How we built it

We built Memory Garden in Unity, integrating the Meta SDK and the Valem Meta Ray-Ban Display Glasses sample UI as our foundation for prototyping for the Meta Ray-Ban AI Display Glasses. For AI, we connected to the OpenAI API. Our 3D assets were customised, modelled on site and modified in Autodesk Maya, then textured in Substance Painter. Photos used in the prototype were sourced from Google, and stories were written by the team and recorded by teammates to simulate loved ones’ voices.

Challenges we ran into

From our developers and technical standpoint

  • Integrating the MR headset and sample UI was more difficult than expected.
  • Micro-gesture recognition and interaction design were delicate challneges.
  • On the co-labs side, we wrestled with merge conflicts with Unity Version across workstations.

From a design and ethics standpoint

  • Translating CST and RT principles into an MR + AI experience forced us to slow down and prioritise safety and comfort over flashy features.
  • We had to continually have the people in mind as the centre of the design process.
  • Given the sensitivity of our user group, ethical and moral questions were discussed and considered with every interaction.

Accomplishments that we're proud of

We’re proud that we built a working prototype that reflects the core design concept and idea. We were able to put all of our separate work together and combine it into a meaningful experience.

What we learned

Designing for a hackathon is super fun and intense! We also came away with a stronger appreciation for ethical questions around AI and XR design in general.

What's next for Memory Garden

It could be cool to implement and test the experience on the Meta Ray-Ban AI Display glasses and other MR to validate comfort, reliability, and accessibility.

  • Collaborating with clinicians, caregivers, and patient advocacy groups to refine the flow and design concept.
  • Running studies to explore how the experience impacts the user’s
  • Building secure, privacy-conscious pipelines so families can safely upload photos, record stories, and manage their data with granular control.

Our long-term vision is to create awareness of designing for good with ethical and changemaking in mind to improve human quality of life.

Ultimately, we hope Memory Garden can help families feel they are walking through this difficult chapter together.

Built With

Share this project:

Updates