Inspiration
Memory loss doesn’t happen all at once. It starts with small moments — forgetting a name, repeating a question, getting confused in a familiar place. Over time, these moments can become more serious, sometimes making loved ones feel like strangers.
We wanted to explore a simple question: What if technology could help bring a little familiarity back into those moments?
That idea led us to build MIRA — Memory & Interaction Recognition Assistant, an AI-powered memory companion designed to provide real-time contextual and emotional assistance for people living with memory-related conditions.
What it does
MIRA (Memory & Interaction Recognition Assistant) is an AI-powered memory companion designed to support people living with memory-related conditions. It uses Google Gemini’s multimodal capabilities to understand live audio and visual context and connect it with personalized family memories.
Family members can provide photos, videos, voice recordings, relationship details, and personal notes. When MIRA recognizes a familiar person or situation, it can provide relevant context and gentle conversational assistance — helping turn moments of confusion into moments of familiarity.
How we built it
MIRA uses Google Gemini’s multimodal capabilities to understand live video and audio input. Family members can provide meaningful context such as photographs, voice recordings, videos, relationship details, and personal notes.
When MIRA observes a familiar person or situation, it connects the current context with the stored family information and can provide a gentle, natural response.
For example, instead of simply saying “Face recognized,” MIRA can provide meaningful context such as:
“This is your daughter Sarah. You last met her last Sunday. You were talking about her dog, Max.”
The goal is not to replace human interaction, but to support it when memory creates a barrier.
We also designed the interface around zero cognitive load — minimal controls, large readable elements, voice-first interaction, and calm visual feedback.
Our future roadmap includes features such as caregiver dashboards, multilingual support, smart-glasses integration, and geofencing alerts that can notify caregivers if a patient moves outside a predefined safe area.
Challenges we ran into
One of our biggest challenges was figuring out how to combine real-time multimodal understanding with personal memory context while keeping the interaction simple enough for someone who may already be experiencing confusion.
We also had to think carefully about privacy and responsible AI. Personal memories are highly sensitive, so user consent, caregiver control, and responsible handling of personal information are essential to the future development of MIRA.
Accomplishments that we're proud of
- Built a working prototype using Google Gemini’s multimodal AI capabilities.
- Integrated real-time visual and audio understanding into the experience.
- Created a personalized memory context using family-provided information.
- Designed a zero-cognitive-load, voice-first interface specifically around the needs of the target users.
- Demonstrated how AI can move beyond simple question-answering into context-aware and emotionally supportive interactions.
- Developed a clear roadmap for extending MIRA into wearable devices, caregiver tools, and safety features.
What we learned
Through this project, we learned that building an AI application isn’t just about making the model smarter. The real challenge is designing how that intelligence interacts with people.
Google Gemini helped us explore how multimodal AI and large-context reasoning can move beyond traditional chatbots and become part of a more human-centered experience.
For us, MIRA is ultimately about one thing:
When memory fades, connection shouldn’t.
What's next for MIRA — Memory & Interaction Recognition Assistant
MIRA is currently a prototype, but our vision is to make it a more complete caregiving companion.
Our next steps are:
- Smart-glasses integration for completely hands-free assistance.
- Geofencing & safety alerts — caregivers are notified when a patient crosses a predefined safe boundary.
- Caregiver dashboard for updates, alerts, and personalized patient context.
- Multilingual support to make MIRA accessible across different languages and families.
- Better personalization through richer family memories and caregiver-provided context.
- Stronger privacy controls, including consent-based memory management and secure handling of sensitive personal information.
Log in or sign up for Devpost to join the conversation.