Spring AI Training Course

1. Introduction to Spring AI

• Setting up projects and configuration
• The function of prompts and their submission
• Developing an initial test
• Selecting a model
• Configuring the model
• An overview of Spring AI features

2. Interpreting responses

• Verifying the relevance of answers
• Evaluating runtime accuracy

3. Prompting in depth

• Utilising prompt templates
• Creating a new prompt template
• Comprehending context
• The significance of the role
• Guiding response generation via options
• Streaming and formatting output
• Response metadata

4. Leveraging your data and documents

• Concepts of RAG (Retrieval-Augmented Generation)
• Configuring the vector store and ingesting documents
• An initial RAG implementation
• Implementing RAG with an advisor
• Modular RAG functionalities

5. The significance of memory in AI

• The necessity of memory
• Integrating and configuring memory for conversations
• Managing the conversation ID
• Enabling persistent memory
• Storing chat memory within a vector store

6. AI Tools

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

7. The Model Context Protocol (MCP)

• The rationale for MCP
• Interacting with an MCP Client
• Developing an MCP Server
• Databases and tools for the MCP Server
• Understanding HTTP and SSE (Server-Sent Events) transport
• Exposing prompts and resources

8. Operational monitoring

• Activating actuator metrics
• Reviewing vector store operations
• Analysing model interactions
• Token counting
• Aggregating data in Prometheus and building dashboards
• Tracing AI operations

9. Safeguarding generative AI

• Regulating documents accessed via RAG
• Securing tools
• Mitigating adversarial prompting
• Moderating user input

10. Common generative patterns

• Content summarisation
• Message translation
• Sentiment analysis

11. The role of Agents

• Defining an agent
• Implementing agentic workflows
• Chaining prompts, task routing, and parallelisation
• Agent access via MCP