Expanding Access to Reproductive Health Advice via Voice-Enabled AI on AWS

“The Melio team is great at what they do - very passionate and skilled. It's been a pleasure working with them.”
Thato Schermer - CEO & Co-Founder of Zoie Health
Case Study
Industry
HealthTech
Company Size
Startup
Problem Space
Reproductive health access
Digital inclusion
Low-connectivity environments
AI voice interfaces
Technology Used
Amazon Bedrock
Amazon Connect
Amazon Lex
AWS Lambda
Amazon DynamoDB
Amazon API Gateway
Business Context
Millions across Africa lack internet access or smartphones, making it difficult to receive sexual and reproductive health information. Zoie Health partnered with Melio AI to build Oobat AI — a voice-enabled chatbot powered by GenAI and delivered via USSD and phone. This MVP aimed to validate engagement and reach within underserved communities.
Watch the Demo
Demo Video - Oobat AI by Melio AdminChallenge
Zoie needed a safe, accessible, and scalable solution to:
- Deliver trustworthy medical guidance to users with no smartphone or internet access
- Provide voice-based interaction through USSD and call-back flows
- Ensure accurate responses based on regional health regulations
- Manage edge cases with human fallback and ongoing evaluation
Solution
Voice-enabled GenAI agent accessible via USSD
The MVP platform integrated with Africa’s Talking USSD service to queue user queries. Human operators could then trigger a call-back via Amazon Connect and transfer to the Oobat AI agent. Key features included:
- Conversational AI via Amazon Lex and Bedrock
- Voice processing with Amazon Connect
- Backend orchestration using Lambda and API Gateway
- User/session data tracked securely in DynamoDB
A custom dashboard allowed testers to mark responses and gather user feedback. Two datasets (knowledge base and evaluation set) ensured safe, relevant medical responses.
A safer, inclusive way to access female health guidance
Oobat AI succeeded in offering real-time, voice-delivered health advice to communities with no internet access. It created a path to support underserved users with privacy, dignity, and medical accuracy.
📞
Voice-enabled AI responses with fallback to human support
🌍
Expanded reach to low-connectivity rural areas in Africa
🗣️
Supports USSD initiation and future multilingual voice integration
🧠
Built-in safety using vetted knowledge and ongoing evaluation sets