MediCall

Your voice powered path to better care

Comment

🩺 MediCall — Voice AI That Makes Insurance-Friendly Healthcare Effortless

MediCall is a multilingual, voice-first AI assistant that makes healthcare access feel like a conversation. Patients can call in and get help with:

  • Understanding their insurance (copay, deductible, coinsurance)
  • Finding nearby in-network hospitals
  • Booking appointments based on real-time availability
  • All powered by natural, real-time speech — in the language they’re most comfortable with.

💡 Inspiration

Too many patients delay care simply because the system is hard to navigate — especially when sick, confused about insurance, or facing a language barrier. We asked:

“What if accessing care was as easy as making a phone call — in any language?”

MediCall helps users understand their coverage, find nearby hospitals that accept their plan, and get care — all through a simple, spoken conversation.

🧠 What We Learned

We tackled tough technical problems to make MediCall feel human, inclusive, and helpful:

  • Multilingual AI Communication: We enabled full speech-to-speech conversations in multiple languages using ElevenLabs + GPT-4.
  • Insurance Clarity: Extracted and simplified confusing insurance terms like deductibles, coinsurance, and copays.
  • n8n-Powered Workflow Routing: Built modular workflows in n8n to route between insurance agents, triage flows, and appointment handlers.
  • Speech-to-Speech Pipeline: Used ElevenLabs for natural, multilingual voice synthesis and response — not just speech-to-text.
  • MongoDB Healthcare Modeling: Designed custom schemas for hospitals, plans, appointments, and call logs.
  • Geolocation Matching: Used Google Maps API and OpenCage API to find in-network hospitals near the caller.
  • Secure Access: Implemented Auth0 to manage patient, provider, and admin roles.
  • Dashboard UX: Built a React + MongoDB dashboard for hospitals to manage bookings and patient interactions.

🌍 Multilingual Support

MediCall supports multiple languages natively — helping patients:

  • Explain symptoms in their native language
  • Understand complex insurance details without jargon
  • Get care without needing a translator or second phone call

This feature is critical for immigrant, elderly, and underserved communities, giving everyone equal access to care.

🤖 Agent Overview

Our n8n-based system is modular, with agents that specialize in key parts of the conversation:

Agent Name Role & Function
UltimateAssistant Routes the conversation based on user intent
insuranceAgent Extracts and explains insurance coverage, finds in-network options
knowledgeAgent Reads/writes data from MongoDB for patients and providers
triageAgent Understands symptoms and classifies urgency of care
appointmentAgent Books and manages appointments with real-time availability checks

🏗️ System Architecture

Layer Tech Used
Speech Interface ElevenLabs (multilingual speech-to-speech)
Conversational AI GPT-4 + n8n agent workflows
Workflow Orchestration n8n (modular logic + conditional routing)
Database MongoDB Atlas
Auth Auth0
Geolocation Google Maps API + OpenCage API
Admin Dashboard React + MongoDB

🔗 Seamless Insurance-to-Care Pipeline

MediCall walks each user through a guided process:

  1. Speaks with them in their preferred language
  2. Understands and explains insurance coverage (copay, deductible, coinsurance)
  3. Finds in-network hospitals nearby
  4. Triages symptoms
  5. Books appointments with available doctors
  6. Sends reminders or updates via SMS/email — powered by n8n

All data is securely stored in MongoDB and visible to providers via the admin dashboard.

🧗 Challenges We Faced

  • Multilingual Accuracy: Ensuring voice output sounds natural and empathetic in each language.
  • Unpredictable User Input: People describe symptoms and insurance differently; we had to design flexible, fallback-ready prompts.
  • Real-Time Voice UX: Latency between speech transcription and generation was tricky to optimize.
  • Workflow Complexity in n8n: Coordinating agents and handling mid-call context required creative logic blocks and switches.
  • Healthcare Data Modeling: Building schemas that reflect real-world insurance coverage and hospital networks.

🚀 Final Thoughts

MediCall is more than a voice assistant — it’s an inclusive, multilingual, AI-powered healthcare gateway. By breaking language and insurance barriers, we bring care to people who need it most — through the universal interface of human conversation.

Built With

Share this project:

Updates