What is Heva - Medical Tourism?
This project involved designing and building a multilingual, AI-powered conversational agent for Heva, an AI-native healthcare platform focused on cross-border patient engagement. Over six months, the agent was developed using LangGraph and integrated with Claude 3.5 and GPT-4o to automate inquiries related to treatments, appointment scheduling, costs, and travel logistics. It served over 1,000 international patients post-launch, increasing appointment conversion rates by up to 45% and reducing provider workload by approximately 30%. The system ensured compliance, real-time data accuracy, and scalability across time zones, reinforcing Heva’s mission of delivering “Care Without Borders.”
Production URL:
heva – Care Without Borders | AI-Native Healthcare Platform
Technology:
ReactJS, Python, AI – ML
List of App Features
Multi-Turn Dialogue Management
Conversation state is tracked through LangGraph memory, enabling smooth transitions and contextual follow-ups throughout the interaction.
Model-Based Dynamic Routing
Claude 3.5 is used for maintaining a conversational tone, while GPT-4o handles complex or fact-based queries to ensure accurate and context-aware responses.
API-Driven Real-Time Actions
Check doctor availability, book consultations, and fetch procedure details, costs, and travel logistics — all powered through real-time API integrations.
Prompt Engineering & Safety Controls
Structured prompt templates ensure consistent responses, while safety filters prevent the agent from giving medical advice. Fallback handling and escalation workflows manage uncertain queries smoothly and safely.
Multilingual Support
English and Spanish out-of-the-box using GPT-4o’s multilingual capability
Scalability Features
Auto-scaling with Kubernetes ensures the system can handle peak traffic effortlessly. Caching optimizes performance by speeding up responses for common queries, while the Streaming API delivers a real-time typing experience for smoother, more engaging interactions.
Monitoring & Continuous Improvement
LangSmith is used for detailed logging and accuracy tracking, ensuring the system remains reliable over time. Additionally, weekly transcript reviews are conducted to continuously refine responses, improve conversational quality, and identify new enhancement opportunities.
Compliance & Privacy Handling
A stateless system design is used to ensure PHI (Protected Health Information) safety by avoiding the storage of sensitive user data. Additionally, medically sensitive topics are automatically redirected to appropriate healthcare professionals or verified resources to maintain compliance and safeguard users.
User Experience Enhancements
Includes human-like typing animation for a natural feel, a friendly and empathetic tone in responses, quick response latency to keep conversations smooth, and a multilingual toggle for seamless switching between supported languages.
Real examples of tech making business better
Let’s Build Something Super
Great ideas need the right tech heroes to back them Fill the form and let our super devs make it real
Want to power up your project?Team up with IT Path Solutions and schedule your power call today