Inspiration
One of our team members shared a deeply personal experience that became the spark for this project. After moving to a new country as an international student, she faced intense loneliness and anxiety amplified by the stress. In our toughest moments, everyone wants to talk to their mom. But she held back, not wanting to worry her. That’s when a thought struck: What if we could feel like we are talking to loved ones -hear their voice, their tone, their comfort, without actually needing to call? That feeling alone brought a sense of calm. We realized how many people silently carry the weight of mental health struggles, simply wishing for the comfort of a familiar voice. And that’s what inspired us to build this app a safe, AI-powered space where healing conversations don’t feel clinical, but personal. Where support feels like home. That made us wonder: what if healing conversations didn’t have to feel like therapy, but instead felt like home?
What it does
Healer.ai is built for emotional comfort and mental healing through familiar, human-like conversation.
The user has two options. They can add a person they want to talk to by uploading a short voice sample (less than a minute), along with details like how that person speaks, what they call the user, and their tone or personality. Our system then clones that voice and creates a personal AI agent.
From there, the user can engage in live, real-time conversations with the person (AI agent) no scripts, no breaks but just flowing dialogue that sounds and feels exactly like the person they miss. The goal is not to simulate therapy but to offer emotional support that feels human and culturally familiar.
How we built it
Challenges we ran into
Achieving low-latency, two-way conversations with natural pauses and no interruptions was technically demanding. We had to balance real-time responsiveness with emotional timing.
ElevenLabs offered excellent voice quality, but initially lacked memory or contextual continuity. We solved this by integrating it with Claude to bring emotional understanding and conversation flow.
We tried incorporating video using tools like Sora and Gemini’s video APIs, but due to restrictions and instability, we decided to leave this for a future version.
Accomplishments that we're proud of
Creating a voice interaction that feels emotionally grounding
Giving users the ability to build something that feels personal and safe
We got positive feedback from two mentors who said this was one of the most meaningful and emotionally grounded projects they’d seen at the hackathon. They appreciated how it tackled mental wellness with empathy and cultural awareness
What we learned
Technology can support emotional well-being, but only when it feels natural and personal
Small cues like tone, timing, and language build deeper trust
People feel safer opening up when they feel understood especially in their own cultural and emotional language
Real-time voice interaction requires precision but is worth the effort for connection
What's next for Healer.ai
We’re just getting started. Here’s what we plan to build next:
Add multilingual support for regional languages and dialects.
Introduce real-time synchronized video avatars to create a more immersive, face-to-face experience during conversations.
Expand personality customization, allowing users to fine-tune tone, warmth, and conversation style.
Build deeper memory retention to help the AI remember past sessions and grow with the user emotionally.
Explore partnerships with mental health professionals to bring this technology into real therapeutic environments.
Built With
- claude
- elevenlabs
- firebase
- git
- javascript
- json
- nextjs
- react
- typescript
- vapi
- vercel
Log in or sign up for Devpost to join the conversation.