Milus.AI

Inspiration

We wanted to tackle a growing problem: loneliness among older adults. Families are often spread out, and technology hasn’t kept up with real human connection. We envisioned a companion that remembers, cares, and bridges that gap.

What it does

Milus.AI is a voice-first AI companion for older adults that provides daily conversation, proactive check-ins, memory recall, and emotional support. Caregivers get dashboards with mood trends, alerts, and weekly summaries.

How we built it

We used Next.js and TypeScript for a fast web interface, Tailwind CSS for responsive design, Featherless.ai for LLM-based memory and conversation, and ElevenLabs for consistent voice responses. The frontend integrates mock APIs for rapid iteration.

Challenges we ran into

Time was tight, only 6 hours. We also had to design voice-first interactions that feel natural, handle privacy for sensitive conversations, and create a caregiver dashboard that is both informative and simple.

Accomplishments that we're proud of

We delivered a fully interactive MVP with voice chat, memory recall, mode-switching, and a caregiver dashboard. The interface is intuitive, the AI “remembers” user info, and the whole system demonstrates real-world social impact.

What we learned

Building for older adults taught us the importance of simplicity, empathy, and privacy. We also learned how to rapidly prototype AI-driven features with mock APIs while keeping the interface polished.

What's next for Milus.AI

We plan to connect Milus to a real backend, expand memory persistence, integrate real-time sentiment detection, and explore proactive suggestions to strengthen real human relationships.

Built With

• featherless.ai
• turbopuffer
• typescript