An AI-powered diabetes chatbot assistant that provides accurate, evidence-based information about diabetes using trusted American Diabetes Association (ADA) resources, powered by AWS Bedrock and Retrieval Augmented Generation (RAG).
Watch the complete demonstration of ADA Clara:
Click the image above to watch the demo
[PLACEHOLDER] Please provide a demo video thumbnail image and save it as
docs/media/demo-thumbnail.png, and update the video URL link above.
| Description | Link |
|---|---|
| Overview | Overview |
| Architecture | Architecture |
| Detailed Architecture | Detailed Architecture |
| Deployment | Deployment |
| User Guide | User Guide |
| API Documentation | API Documentation |
| Infrastructure | Infrastructure |
| Modification Guide | Modification Guide |
| Credits | Credits |
| License | License |
This application combines AI-powered conversational AI with intelligent knowledge retrieval to deliver accurate, evidence-based diabetes information. Built on a serverless architecture with RAG (Retrieval Augmented Generation), automated content management, and comprehensive analytics, ADA Clara enables healthcare organizations to provide 24/7 diabetes support with trusted ADA resources.
- AI-Powered Chatbot powered by AWS Bedrock with Claude Sonnet 3
- RAG System using Amazon Bedrock Knowledge Base with Titan Text Embedding V2 for vector embeddings
- Multi-Language Support with language selection and interface localization
- Automated Knowledge Base with web scraping from diabetes.org
- Admin Dashboard with real-time analytics and conversation insights
- Escalation Management for connecting users with healthcare professionals
- Source Citations with links to original diabetes.org content
- Question Analytics tracking frequently asked and unanswered questions
The application implements a serverless, event-driven architecture with a RAG-powered AI system at its core, combining automated content processing with intelligent question answering and comprehensive analytics.
For a detailed deep dive into the architecture, including core principles, component interactions, data flow, security, and implementation details, see docs/architectureDeepDive.md.
For a detailed overview of the user journey and application workflow, including step-by-step user interactions, see docs/userGuide.md.
Deploying ADA Clara is simple and requires no local dependencies. Everything can be done from AWS CloudShell:
- Open AWS Console and start CloudShell
- Clone the repository:
git clone https://github.com/ASUCICREPO/ADA-Clara.git - Navigate to the project:
cd ADA-Clara - Make the script executable:
chmod +x deploy.sh - Run the deployment:
./deploy.sh
The deployment script handles everything automatically, including backend infrastructure, frontend deployment via CodeBuild, and knowledge base setup. For detailed instructions, see docs/deploymentGuide.md.
For detailed backend testing and usage instructions, including configuration steps and how to test the application from AWS Console, see docs/userGuide.md.
For frontend user guide and application features, see docs/userGuide.md.
For a detailed overview of the application infrastructure, including component interactions, AWS services, and data flow, see docs/architectureDeepDive.md.
- API Documentation - Comprehensive API reference for all endpoints
- Architecture Deep Dive - Detailed system architecture and design
Steps to implement optional modifications such as changing the Bedrock model, adding new features, or customizing the frontend can be found here.
This application was architected and developed by Shaashvat Mittal, Sean Sannier, and Omdevsinh Zala with solutions architect Arun Arunachalam, program manager Thomas Orr and product manager Rachel Hayden. Thanks to the ASU Cloud Innovation Center Technical and Project Management teams for their guidance and support.
See LICENSE file for details.