How aws-solutions/qnabot-on-aws works
Main Components
a. User Interface Layer:
• Content Designer UI (Admin Interface)
• Chat Client UI
• Amazon Alexa Interface
• Amazon Connect Interface
• Genesys Interface
b. Authentication Layer:
• Amazon Cognito for user authentication
• IAM roles and permissions
c. API Layer:
• Amazon API Gateway for REST endpoints
• WebSocket connections for real-time chat
d. Core Processing Layer:
• Bot Fulfillment Lambda
• Lex Build Lambda
• Import/Export Lambda functions
• Schema validation Lambda
• Custom Lambda hooks
e. Storage Layer:
• Amazon OpenSearch Service (question bank)
• Amazon DynamoDB (session management)
• Amazon S3 (assets and configurations)
• AWS Systems Manager Parameter Store (settings)
f. AI/ML Services Layer:
• Amazon Lex
• Amazon Bedrock
• Amazon Kendra
• Amazon Comprehend
• Amazon Translate
• Amazon SageMaker (optional)
g. Monitoring & Logging Layer:
• Amazon CloudWatch
• Kinesis Data Firehose
• OpenSearch Dashboards
Important Workflows to Highlight
a. Content Management Flow:
• Admin → Content Designer UI → API Gateway → Lambda → OpenSearch
• Import/Export processes through S3
b. Chat Interaction Flow:
• User → Chat Interface → Lex → Bot Fulfillment Lambda → OpenSearch/Kendra
• Multi-language support through Amazon Translate
• LLM integration through Bedrock/SageMaker
c. Monitoring Flow:
• Lambda → CloudWatch → Kinesis → S3
• OpenSearch Dashboards for analytics
Version Information
• Current version number
• Supported AWS regions
• Required service versions
Remember to include the optional components (like Kendra and SageMaker) with clear indicators that they are optional, and show the alternative paths when these services are not used.
This system design should help developers understand both the overall architecture and the specific implementation details of the QnABot solution.