View the changelog

Want to know the history of this service? View the change logs

Contact the author

Any questions? Feel free to contact the owners

Receives 7 messages

This service receives messages from other services

Overview

The Chat Service is a specialized bounded context within the BookWorm ecosystem that implements natural language processing capabilities to provide interactive conversations with users, following our AI Integration Strategy. This service is responsible for:

• Processing user chat prompts and generating contextually relevant responses
• Maintaining chat session state and history
• Providing book recommendations and information through conversational interfaces
• Leveraging large language models (LLMs) to understand and respond to user queries
• Integrating with the catalog service to access book metadata and information

The domain model follows strategic DDD patterns with clearly defined aggregates, entities, and value objects. The Chat domain enforces business rules such as prompt validation, session management, and response generation.

Key Capabilities

• Conversational AI: Natural language understanding and generation using state-of-the-art LLMs
• Context Management: Maintains conversation history and context for coherent multi-turn dialogues
• Tool Integration: Leverages Model Context Protocol (MCP) for dynamic tool usage
• Streaming Responses: Real-time streaming of AI responses using SignalR for better user experience

Component Diagram

Core Features

Feature	Description
Create Chat	Initialize a new chat session with a prompt and receive a unique identifier.
Cancel Chat	Terminate an existing chat session.
Conversational AI	Engage with an AI assistant that can answer questions about books and provide recommendations.
Context Awareness	Maintain conversation context across multiple interactions within a session.

Architecture diagram

Infrastructure

The Chat Service is deployed on Microsoft Azure and leverages various Azure services for scalability, reliability, and performance. The infrastructure components include:

Security Considerations

• Authentication: JWT-based authentication with user context isolation
• Rate Limiting: Per-user request throttling to prevent abuse
• Content Filtering: Automated moderation of inappropriate content
• Data Privacy: Conversation history encrypted at rest
• Prompt Injection Protection: Input sanitization and validation
• API Key Management: Secure storage of LLM API credentials in Azure Key Vault

Monitoring and Observability

Key Metrics

• Response Latency: Time to first token and total generation time
• Token Usage: Track token consumption per user and model
• Conversation Quality: User satisfaction and engagement metrics
• Error Rates: Failed generations and timeout occurrences
• Tool Usage: Frequency and success rate of MCP tool calls