Spring AI Training Course

1. Introduction to Spring AI

• Initiating projects and configuration
• The function of prompts and prompt submission
• Creating initial tests
• Selecting a model
• Configuring model parameters
• Overview of Spring AI features

2. Interpreting responses

• Verifying the relevance of answers
• Assessing runtime accuracy

3. Detailed prompt engineering

• Utilizing prompt templates
• Creating custom prompt templates
• Grasping context
• The significance of defining roles
• Guiding response generation via options
• Streaming and formatting output
• Response metadata

4. Utilizing proprietary data and documents

• Comprehending RAG (Retrieval-Augmented Generation)
• Establishing vector stores and ingesting documents
• Implementing a basic RAG workflow
• Implementing RAG with an advisor
• Modular RAG functionalities

5. The importance of memory in AI

• The necessity of memory
• Implementing and configuring memory for conversations
• Managing conversation IDs
• Enabling persistent memory
• Storing chat history in vector stores

6. AI Tools

• Building tool-enabled applications
• Exploring tool capabilities
• Developing and deploying tools
• Using functions as tools

7. The Model Context Protocol (MCP)

• The need for MCP
• Developing an MCP Client
• Creating an MCP Server
• Integrating databases and tools into the MCP Server
• Understanding HTTP and SSE (Server-Sent Events) transport
• Publishing prompts and resources

8. Monitoring operations

• Enabling actuator metrics
• Tracking vector store operations
• Monitoring model interactions
• Token usage analysis
• Aggregating data in Prometheus and building dashboards
• Tracing AI workflows

9. Security in generative AI

• Controlling document access via RAG
• Securing tool usage
• Mitigating adversarial prompting
• Moderating user input

10. Standard generative patterns

• Summarizing content
• Translating messages
• Analyzing sentiment

11. The role of Agents

• Defining agents
• Building agentic workflows
• Chaining prompts, routing tasks, and parallel processing
• Accessing agents through MCP