Inspiration

Mental health is the silent pandemic of our generation. While millions suffer daily from stress, anxiety, or emotional overwhelm, access to therapy remains limited, expensive, or stigmatized. I asked a simple question: What if someone could just talk—and be heard, supported, and healed—anytime, anywhere?

This question led me to build empathi ei — a text-voice-video emotional wellness platform that listens like a friend, supports like a therapist, and grows with you like a journal. The ei in the name is symbolic standing for 'emotional intelligence'.

What it does

Empathi ei is an AI-powered emotional support system that users can talk to in real time, like a non-judgmental, always-available companion. It uses advanced conversational AI and voice synthesis to:

🧠 Listen empathetically and provide support through natural conversations.

🧘 Guide personalized meditations with a calm AI voice.

📍 Log global mood data through our FeelLoop Map, showing collective emotional states.

📊 Offer mood insights and self-awareness tools, visualizing your emotional journey over time.

All in one app — accessible, text-voice-video based, and backed by science.

How I built it

Empathi AI was built entirely on Bolt, using a no-code-first approach to AI prototyping. Key technologies:

ElevenLabs Voice API – for realistic and soothing AI-generated voice agents.

Openrouter’s GPT-based LLM – for nuanced emotional understanding and therapeutic dialogue.

Supabase – for real-time database operations, mood tracking, and feedback storage. Tavus AI-for conversational video interface AI therapist.

Dynamic agents – powered by Bolt's multi-agent orchestration for features like mood journaling, voice meditation, and emotional analysis.

The entire workflow is text-voice-video based, with ethical AI guardrails built in.

Challenges I ran into

Crafting a system prompt that balanced empathy and responsibility — especially in emotional conversations.

Integrating voice agents with memory in a way that felt natural and supportive.

Managing global emotion data in real-time while respecting privacy and scalability.

Creating a multi-agent architecture that still felt unified and seamless to the user.

Accomplishments that I am proud of

Built a fully functional emotional voice assistant in < 7 days using Bolt.

Designed and deployed a global mood tracking system (FeelLoop).

Integrated conversational AI with meditation, journaling, and emotion analysis.

Received highly emotional, real user feedback that validated the core experience.

What I learnt

Voice and video are more than input — it's connection.

People don't always need solutions, they need presence — and my AI agent succeeded in delivering that.

Building ethical AI for mental health demands empathy, safety, and continuous iteration.

No-code platforms like Bolt can power real, impactful AI solutions when used creatively.

What's next for empathi ei

🔒 Adding multilingual, emotion-aware AI agents that can hold memory over time.

📱 Publishing Empathi AI as a public voice-first app for teenagers and working professionals.

❤️ Partnering with mental wellness nonprofits and education platforms to provide voice and video-based support tools.

🧪 Training emotion-specific agents like a Gratitude Coach, Inner Child Helper, and Burnout Listener.

Empathi AI is not just a project. It’s a quiet revolution in emotional wellness. One voice. One emotion. One moment at a time.

Built With

  • 11labs
  • bolt-(no-code-ai-orchestration-platform)
  • openrouter
  • supabase-(as-the-backend-database-and-real-time-logging-service)
  • tavus-ai-(for-dynamic-ai-generated-video-agents)
Share this project:

Updates