Retail Banking Chatbot

A retail banking assistant for AWSome Bank, powered by Amazon Bedrock with RAG-based knowledge retrieval and real-time transaction lookup. Demonstrates the InteractiveAgent construct with Knowledge Base integration, custom tools, and Cognito authentication.

What This Example Demonstrates

• InteractiveAgent with real-time streaming responses (REST API + Server-Sent Events)
• Knowledge Base (RAG) using Amazon Bedrock KB with S3 Vectors for banking FAQ retrieval
• Custom tool (lookup_transactions) querying DynamoDB for customer transaction history
• Cognito authentication with user sign-up/sign-in
• React frontend with banking theme hosted on S3 + CloudFront
• Automated KB provisioning — Knowledge Base, vector store, data source, and ingestion all created in-stack

Architecture

┌─────────────┐     ┌──────────────┐     ┌──────────────────-───┐
│   React UI  │────▶│  CloudFront  │────▶│   API Gateway (REST) │
│  (S3 hosted)│     │     (CDN)    │     │  + Cognito Authorizer│
└─────────────┘     └──────────────┘     └──────────┬────-──────┘
                                                     │
                                                     ▼
                                          ┌────────────────────--------─┐
                                          │  Lambda (InteractiveAgent)  │
                                          │  - Strands Agent Framework  │
                                          │  - Lambda Web Adapter (SSE) │
                                          └──────┬──────┬────--------───┘
                                                 │      │
                              ┌──────────────────┘      └──────────────────┐
                              ▼                                            ▼
                   ┌─────────────────────┐                    ┌──────────────────┐
                   │  Bedrock Claude LLM │                    │   Agent Tools    │
                   │  (reasoning engine) │                    ├──────────────────┤
                   └─────────────────────┘                    │ retrieve_from_   │
                                                              │ knowledge_base() │
                              ┌───────────────────────-─┐     │                  │
                              │  Bedrock Knowledge Base │◀────│ lookup_          │
                              │  + S3 Vectors Index     │     │ transactions()   │
                              │  + Banking FAQ Docs     │     └────────┬─────── ─┘
                              └───────────────────────-─┘               │
                                                                       ▼
                                                              ┌──────────────────┐
                                                              │    DynamoDB      │
                                                              │  (Transactions)  │
                                                              └──────────────────┘

Request Flow

1. User sends a message through the React frontend
2. Request is authenticated via Cognito JWT and routed through API Gateway REST API
3. API Gateway invokes Lambda using response streaming mode (ResponseTransferMode: STREAM) for real-time token-by-token delivery
4. Lambda runs a FastAPI server via Lambda Web Adapter, streaming SSE events back through API Gateway to the frontend
5. The Strands agent framework orchestrates the response:
   • For product/policy questions → calls retrieve_from_knowledge_base tool → queries Bedrock KB
   • For transaction queries → calls lookup_transactions tool → queries DynamoDB
   • For combined questions → uses both tools, then synthesizes a response
6. Response streams back to the frontend via Server-Sent Events (SSE)

Components

Component	Service	Purpose
Frontend	S3 + CloudFront	React chat UI with banking theme
Auth	Cognito User Pool	User sign-up/sign-in with JWT tokens
API	API Gateway REST	Chat endpoint with Cognito authorizer and response streaming mode
Agent	Lambda + Strands	AI agent with tool orchestration, streamed via Lambda Web Adapter
LLM	Bedrock (Claude)	Reasoning, tool selection, response generation
Knowledge Base	Bedrock KB + S3 Vectors	RAG retrieval for banking FAQs
FAQ Storage	S3	Source documents for KB ingestion
Embeddings	Titan Embed Text v2	Converts FAQ text to vectors (1024 dimensions)
Transactions	DynamoDB	Customer transaction history (on-demand billing)
Sessions	S3	Conversation history (24-hour TTL)
Observability	CloudWatch + X-Ray	Logs, metrics, and tracing

Prerequisites

• AWS CLI configured with appropriate credentials
• Node.js 18+
• Python 3.11+
• CDK bootstrapped: npx cdk bootstrap
• Amazon Bedrock model access enabled for Claude and Titan Embed Text v2

Deployment

# From the retail-banking-chatbot directory:

# 1. Build the library (from repo root, if not already done)
cd /path/to/repo/root
SKIP_DOCGEN=true npx projen build
npx projen package:js

# 2. Install and deploy infrastructure
cd examples/chatbot/retail-banking-chatbot/infrastructure
npm install
npx cdk deploy --profile <your-profile> --require-approval never --outputs-file ../outputs.json

# 3. Seed transaction data
cd ..
./seed-data.sh

# 4. Configure frontend (copy values from outputs.json)
cat > frontend/.env.production << EOF
REACT_APP_CHAT_API_ENDPOINT=<ChatApiEndpoint from outputs.json>
REACT_APP_USER_POOL_ID=<UserPoolId from outputs.json>
REACT_APP_USER_POOL_CLIENT_ID=<UserPoolClientId from outputs.json>
REACT_APP_REGION=<Region from outputs.json>
EOF

# 5. Build frontend and redeploy
cd frontend && npm install && npm run build && cd ../infrastructure
npx cdk deploy --profile <your-profile> --require-approval never

Sample Prompts

Product & Policy Questions (KB Retrieval)

• "What types of accounts does AWSome Bank offer?"
• "What are the overdraft fees? Is there any way to avoid them?"
• "Do you offer any credit cards with travel rewards?"
• "What CD rates do you currently offer?"
• "How do I send money using Zelle?"

Transaction Lookup (DynamoDB Tool)

• "Can you show me my recent transactions?" → Agent will ask for customer ID
• "Show me transactions for CUST-001"
• "Look up transactions for CUST-002"

Combined (Tool + KB)

• After viewing CUST-001 transactions: "I didn't make that suspicious purchase. What should I do?"
• After viewing CUST-002 transactions: "What home equity options does AWSome Bank offer?"
• After viewing CUST-003 transactions: "If I open a 12-month CD with $10,000, what would my return be?"

Available Customers: CUST-001 (Sarah Chen), CUST-002 (Marcus Johnson), CUST-003 (Dorothy Williams)

Sample Data

• FAQ Documents (sample-docs/): Banking FAQ covering accounts, fees, transfers, cards, digital banking, loans
• Transaction Data (sample-data/): 34 sample transactions across 3 customer personas

Monitoring

# View agent Lambda logs
aws logs tail /aws/lambda/<AgentFunctionName>-secured --follow --profile <your-profile>

# Check Knowledge Base ingestion status
aws bedrock-agent get-knowledge-base --knowledge-base-id <KnowledgeBaseId> --profile <your-profile>

# Query DynamoDB transactions
aws dynamodb query --table-name <TransactionsTableName> \
  --key-condition-expression "customerId = :id" \
  --expression-attribute-values '{":id":{"S":"CUST-001"}}' \
  --profile <your-profile>

Troubleshooting

Issue	Cause	Fix
Agent not using latest features	Stale library tarball	Run npx projen package:js, reinstall, redeploy
Frontend shows old version	CloudFront cache	Invalidate: aws cloudfront create-invalidation --distribution-id <ID> --paths "/*"
Auth error on frontend	Missing .env.production	Generate from outputs.json and rebuild frontend
Empty transaction results	Table not seeded	Run ./seed-data.sh

Cleanup

cd infrastructure
npx cdk destroy --profile <your-profile>

This removes all resources including the DynamoDB table, Knowledge Base, S3 buckets, Lambda functions, and CloudFront distribution.

Related Constructs

• InteractiveAgent — Real-time streaming agent with auth and session management
• BedrockKnowledgeBase — Knowledge base integration for RAG
• Frontend — Static web hosting with CloudFront