An intelligent AI-powered customer support chatbot that helps US PreK-12 educators (administrators and teachers) get instant answers about PLTW curriculum implementation, product purchasing, training, assessments, payment, rostering, grants, and technical guidance. Built with AWS Bedrock Agent and NextJS.
PLTW.Demo.mp4
| Description | Link |
|---|---|
| Overview | Overview |
| Architecture | Architecture |
| Detailed Architecture | Architecture Deep Dive |
| Deployment | Deployment Guide |
| User Guide | User Guide |
| API Documentation | API Documentation |
| Modification Guide | Modification Guide |
| Credits | Credits |
| License | License |
PLTW Support Assistant (Jordan) is a conversational AI assistant designed to serve as the first line of customer support for US PreK-12 educators, including District and School administrators, CTE directors, and teachers using Project Lead The Way (PLTW) curriculum. Jordan enables educators to get instant, accurate answers to tier 1 questions about curriculum implementation, product purchasing, training, assessments, payment, rostering, grants, and technical guidance through natural language conversations in English and Spanish.
- AI-Powered Conversations using AWS Bedrock Agent with Knowledge Base integration
- Real-time Responses via WebSocket API for a natural chat experience
- Knowledge Base Integration with PLTW documentation (pltw.org, knowledge.pltw.org, curriculum resources)
- Response Citations displaying source links for AI responses
- Bilingual Support (English/Spanish) with automatic translation via AWS Translate
- LLM-Powered Categorization automatically categorizes conversations using Nova Pro
- Escalation Queue with SQS integration for human agent handoff
- Sentiment Analysis scheduled analysis of conversation sentiment
- Feedback Mechanism to collect user satisfaction data (thumbs up/down)
- Admin Dashboard with analytics charts (Nivo), conversation logs, and metrics
- Conversation Detail View for reviewing full conversation history
- Responsive Design optimized for both desktop and mobile devices
The application implements a serverless architecture on AWS, combining:
- Frontend: Next.js application hosted on AWS Amplify
- Backend: AWS CDK-deployed infrastructure with API Gateway (WebSocket + REST) and Lambda
- AI Layer: AWS Bedrock Agent with Knowledge Base for RAG-based question answering, plus Nova Pro for categorization and sentiment analysis
- Data Storage: DynamoDB for conversation history and connections
- Queue: Amazon SQS FIFO queue for escalation handling
- Authentication: Amazon Cognito for admin dashboard access
- Translation: AWS Translate for bilingual support
For a detailed deep dive into the architecture, see docs/architectureDeepDive.md.
For detailed deployment instructions, including prerequisites and step-by-step guides, see docs/deploymentGuide.md.
# Backend
cd backend
npm install
node build-lambda.js
cd cdk
npm install
export BEDROCK_AGENT_ID=your-agent-id
export BEDROCK_AGENT_ALIAS_ID=your-alias-id
npx tsc && npx cdk deploy --all
# Frontend
cd frontend
npm install
npm run buildFor detailed usage instructions with screenshots, see docs/userGuide.md.
For complete API reference including chat endpoints, admin APIs, and user management, see docs/APIDoc.md.
For developers looking to extend or customize this project, see docs/modificationGuide.md.
├── backend/
│ ├── cdk/
│ │ ├── bin/
│ │ │ └── app.ts # CDK app entry point
│ │ └── lib/
│ │ ├── amplify-stack.ts # AWS Amplify hosting stack
│ │ ├── cognito-stack.ts # Cognito authentication stack
│ │ ├── dynamodb-stack.ts # DynamoDB tables stack
│ │ ├── rest-api-stack.ts # REST API Gateway stack
│ │ ├── sqs-stack.ts # SQS escalation queue stack
│ │ └── websocket-stack.ts # WebSocket API Gateway stack
│ ├── src/
│ │ ├── handlers/
│ │ │ ├── rest/
│ │ │ │ ├── admin.ts # Admin dashboard API (metrics, conversations)
│ │ │ │ ├── feedback.ts # Feedback submission handler
│ │ │ │ └── sentiment.ts # Scheduled sentiment analysis
│ │ │ └── websocket/
│ │ │ ├── connect.ts # WebSocket connection handler
│ │ │ ├── disconnect.ts # WebSocket disconnect handler
│ │ │ └── sendMessage.ts # Main chat Lambda
│ │ ├── services/
│ │ │ ├── bedrock.service.ts # Bedrock Agent invocation
│ │ │ ├── categorization.service.ts # LLM-powered conversation categorization
│ │ │ ├── categorization.constants.ts # Category definitions
│ │ │ ├── dynamodb.service.ts # DynamoDB operations
│ │ │ ├── escalation.service.ts # Escalation queue service
│ │ │ ├── sentiment.service.ts # Sentiment analysis service
│ │ │ ├── translate.service.ts # AWS Translate service
│ │ │ └── websocket.service.ts # WebSocket messaging
│ │ └── types/
│ │ └── index.ts # TypeScript type definitions
│ ├── build-lambda.js # Lambda bundler script
│ └── package.json
├── frontend/
│ ├── app/
│ │ ├── admin/
│ │ │ ├── layout.tsx # Admin layout
│ │ │ └── page.tsx # Admin dashboard page
│ │ ├── layout.tsx # Root layout
│ │ ├── page.tsx # Main chatbot page
│ │ └── globals.css # Global styles
│ ├── components/
│ │ ├── admin/
│ │ │ └── ConversationDetailModal.tsx # Conversation detail view
│ │ ├── charts/
│ │ │ ├── ConversationVolumeChart.tsx # Volume over time chart
│ │ │ ├── EscalationReasonsChart.tsx # Escalation reasons pie chart
│ │ │ ├── TopCategoriesChart.tsx # Top categories bar chart
│ │ │ └── UserSatisfactionChart.tsx # Satisfaction pie chart
│ │ ├── ChatWindow.tsx # Main chatbot interface
│ │ └── CitationDisplay.tsx # Source citation display
│ ├── config/
│ │ └── translations.ts # Bilingual translations
│ ├── contexts/
│ │ ├── AuthContext.tsx # Authentication context
│ │ └── LanguageContext.tsx # Language selection context
│ ├── hooks/
│ │ └── useWebSocket.ts # WebSocket connection hook
│ ├── lib/
│ │ ├── amplify-config.ts # AWS Amplify configuration
│ │ └── api.ts # API client utilities
│ ├── public/ # Static assets (logos, icons)
│ └── package.json
├── docs/
│ ├── architectureDeepDive.md # Detailed architecture documentation
│ ├── deploymentGuide.md # Deployment instructions
│ ├── userGuide.md # User guide
│ ├── APIDoc.md # API documentation
│ ├── modificationGuide.md # Customization guide
│ └── media/
│ └── user-interface.gif # Demo video
├── README.md # This file
└── LICENSE # MIT License
BEDROCK_AGENT_ID- Bedrock Agent IDBEDROCK_AGENT_ALIAS_ID- Bedrock Agent Alias IDNOVA_PRO_MODEL_ID- Nova Pro model ID (default:amazon.nova-pro-v1:0)
NEXT_PUBLIC_WEBSOCKET_URL- WebSocket API endpointNEXT_PUBLIC_API_URL- REST API endpointNEXT_PUBLIC_USER_POOL_ID- Cognito User Pool IDNEXT_PUBLIC_USER_POOL_CLIENT_ID- Cognito Client ID
This application was architected and developed by Pratiksha Wadibhasme, Sreeram Sreedhar, and Omdevsinh Zala, with solutions architect Arun Arunachalam, program manager Thomas Orr and product manager Rachel Hayden. Thanks to the ASU Cloud Innovation Centre Technical and Project Management teams for their guidance and support.
This project is distributed under the MIT License.