AI Agents & Agentic Workflows with Spring AI, MCP and Java

Name: AI Agents & Agentic Workflows with Spring AI, MCP and Java
Rating: 4.6 (34 reviews)

Model Context Protocol, Spring Boot, MCP Servers, ChatClient, ChatMemory, OpenAI, Gemini, Ollama, Integration Testing

Role Play

Created byVinoth Selvaraj

Last updated 6/2026

English

English [Auto],

What you'll learn

Build AI Agents using Spring AI and Java
Design Agentic Workflows and multi-turn reasoning systems
Implement MCP Servers using Spring Boot
Build and expose MCP Tools, Resources and Prompts
Integrate OpenAI, Gemini and local LLMs using Ollama
Build Human-in-the-Loop workflows using Elicitation
Handle asynchronous workflows using Progress Notifications
Write Integration Tests for MCP-based AI systems
Implement Structured Output and Prompt Engineering techniques
Use ChatClient, ChatMemory and Advisors effectively

Course content

17 sections • 190 lectures • 11h 53m total length

What You Will Learn5:30

Introduction to Generative AI5:16
Understanding LLMs & Core Parameters2:25
Introduction to Ollama3:14
Explore how Ollama enables local model execution on your machine with a docker-like workflow. Install, download models, and run via commands, while noting hardware and API considerations.
*** Source Code & GitHub Project Repository ***0:05
Running Ollama Using Docker2:45
Running Models Locally with Ollama4:59
Creating an OpenAI API Key8:29
Creating a Gemini API Key3:14
Create a Gemini API key in Google AI studio by creating a project and copying the key. Explore free-tier limits, billing, and quota options, including models like 3.1 flashlight.

Introduction to ChatClient4:05
LLMs Are Non-Deterministic3:49
Playground Project Structure Discussion5:06
Playground Project Setup8:25
First LLM Interaction3:51
Learning Approach: Start Simple1:15
First LLM Interaction Demo4:16
Clarification: Which Spring Profile to Use1:14
Implementing Streaming Responses4:53
Clarification: Why We Use JUnit for Demos1:00
LLM Tokens Overview3:31
OpenAI Tokenizer Demo1:27
Understanding Metadata, Tokens, and Rate Limits6:51
OPTIONAL: Max Tokens6:31
OPTIONAL: Fine-Tuning Creativity with Temperature3:25
Configuring Default ChatOptions0:41
Dynamic Model Selection at Runtime5:12
Configuring Multi-Model Beans3:14
Working with Multi-Model Providers5:37
Native Structured Output6:04
Structured Output: Single Entity Mapping3:34
Structured Output: Multiple Entity Mapping2:59
Multiple Entity Mapping: OpenAI Issue and Workaround2:54
Core Concept: System Prompts2:47
System Prompt Implementation Demo3:30
Default System Prompts0:52
Dynamic Context with Prompt Templates3:20
Prompt Templates Demo1:56
COURSE ASSIGNMENT: Movie Recommender Engine
Troubleshooting the Movie Recommendation Issue1:21
Prompt Engineering: Iterative Refinement2:14
Movie Recommendation Issue Fix1:06
Section Summary & Key Takeaways6:27

TalentFlow: Application Overview4:43
Multi-Module Project Structure Overview2:33
Exploring Candidate Service5:10
Explore a candidate service with a one-to-many candidate and work experience schema, exposing a minimal api that returns candidate data by id via a dto mapper.
Exploring Job Service6:42
Career Advisor: Architecture Discussion4:32
Career Advisor: Project Setup2:16
Implementing Job and Candidate Clients5:47
Career Advisor Prompts3:59
Clarification: Do we have duplicate tasks?1:58
Find Jobs API Implementation: Part 16:50
Implement a FindJobs API that fetches candidate and job data, scores matches with an LLM, and provides justified reasoning via a career advisor client.
Find Jobs API Implementation: Part 23:16
Implement a find jobs api by injecting job, candidate, and advisor clients, retrieving candidate skills, searching jobs by skills, evaluating with the advisor, and returning a sorted match score list.
Compare Jobs API Implementation4:15
Create a compare jobs API that fetches candidate and job details, uses an LLM to compare based on skills, location, and salary, and returns a best job recommendation.
Generate Resume API Implementation4:15
Career Advisor Configuration8:13
Career Advisor: Managing Secrets2:02
Career Advisor APIs Demo7:22
Exploring Hiring Service4:03
Hiring Service: Job Client Implementation0:57
Hiring Service Prompts4:25
Hiring Advisor Implementation4:38
Job Application Submission: Architecture & Design Discussion2:46
Implement the job application submission API, persist the application via the service and DTO mapper, and trigger an asynchronous event to review with an LLM, updating match score and reasoning.
Job Application Submission Implementation5:04
Job Application Evaluation Implementation3:43
Fetching Job Applications API Implementation1:47
Fetching Candidate Applications API Implementation6:18
Hiring Service Configuration2:27
Build and configure the hiring advisor client using a chat client builder, Jackson JSON mapper, and resource loader, loading prompt templates and enabling native structured output for requests.
Hiring Service APIs Demo3:45
TalentFlow: Environment Preparation1:26
TalentFlow: End-to-End Demo10:11
Section Summary & Key Takeaways1:30
Enhance an existing microservices app by layering AI on top of candidate, job, and hiring services, providing guidance for candidates and insights for recruiters.

The Problem: Why We Need the Model Context Protocol (MCP)3:26
Setting Expectations: Core Protocol Theory1:22
Protecting Data in MCP1:29
High-Level MCP Architecture Overview4:47
Core Capabilities: Understanding MCP Server Features1:33
MCP Capability Discovery3:30
Request-Response Lifecycle: MCP Execution Flow4:50
Clarification: How LLM Executes Tools2:18
Understanding MCP Transports3:04
Spring MCP Modules1:13
MCP Playground: Project Structure Discussion2:56
Explore a Spring Boot Maven playground for the MCP, wiring MCP server and host with external LLMs like Google Gen AI or OpenAI, and run independent sectioned hosts and servers.
MCP Playground: Project Setup7:27
NOTE: Model Support For Tool Calling1:24
MCP Properties & Secrets2:20
Developing Our First MCP Tool2:43
[OPTIONAL] Introduction To Anthropic MCP Inspector3:08
[OPTIONAL] Testing Tool Calls with MCP Inspector3:21
[OPTIONAL] Best Practices: Tool Naming and Descriptions3:57
Architectural Discussion: Host-Server Tool Integration & Execution1:59
MCP Host Project Setup7:52
System Message For Tool Calling0:58
End-to-End MCP Tool Call Demo5:19
Multi-Tool Execution Demo1:03
How LLM Executes Multiple Tool Calls2:02
Clarification: Tool Execution Response Void Type1:17
Looking Ahead: The Role of MCP Resources and Prompts1:33
Section Summary & Key Takeaways4:36

Section Goal2:23
Best Practices: Tool Parameter Name and Description1:47
File Operations Project Setup3:41
File Operations Demo6:51
Explore file operations with the file tools, including creating, reading, deleting, converting csv to json, merging files, and managing runtime directories under target/classes.
Structured Input From LLM0:59
Learn how to structure tool parameters using a Java record to pass a single structured input, describe each field, and keep code clean while scaling parameter handling.
Providing Context to AI Models via System Messages3:22
Structured Input: Practical Demo3:02
Tool Design Approaches1:33
Pattern 1: Assistant-Driven Decisions5:47
Assistant-Driven Decisions: Vulnerability Demo4:25
Architectural Pitfalls: Why Assistant-Driven Decisions Fail1:12
Communicating Tool Execution Results1:51
Pattern 2: Encapsulated Tool Design3:28
Handling Tool Execution Errors Robustly2:21
Capture tool errors as part of the tool response, avoid exceptions, and use clear messages with status codes such as 400, 401, 404 in MCP-hosted LLM interactions.
Pattern 3: Encapsulated & Robust Tool Design1:38
Advanced Context: Passing Metadata to MCP Servers via Meta3:04
MCP Meta: Project Setup5:55
MCP Meta: Demo2:22
Advanced Error Handling via Data-Oriented Programming1:41
Section Summary & Key Takeaways3:35

Introduction to Progress Notifications3:11
Core Concept: Understanding Progress Tokens2:00
How Progress Notifications Work Between Host & Server3:55
Real-Time Streaming: Exposing Progress Updates via Flux7:24
Expose real-time progress updates to the UI using a Sync object and Flux, broadcasting via multicast with backpressure in reactive programming.
How Progress Notifications Work Between Host & UI2:42
Create a unique progress token to route per-user updates from the MCP host to the UI, with a flux API filtering by token.
Progress Notification: Project Setup8:24
End-to-End Progress Notification Demo2:21

Requirements

No prior AI knowledge is required. We will start from the fundamentals with a hands-on approach.
Knowledge of Java and Spring Boot is required.
OpenAI and Gemini APIs may incur small usage costs. Expected cost for this course is approximately 1 USD.

Description

Build AI Agents and Agentic Workflows using Spring AI, MCP and Java. (Latest Spring Boot 4.0, Spring AI 2.0)

This entire course is about developing our own AI Agents From Scratch. It is a deep-dive, architecture-first masterclass on building production-grade AI Agents and Agentic Workflows using Java, Spring AI and the Model Context Protocol (MCP).

What you will master:

Building AI Agents using Spring AI and Java
Designing Agentic Workflows and multi-turn reasoning systems
Understanding MCP Architecture and communication flow
Implementing MCP Tools, Resources and Prompts
Building Human-in-the-Loop workflows using Elicitation
Handling asynchronous workflows using Progress Notifications
Integrating OpenAI, Gemini and local models using Ollama
Using ChatClient, ChatMemory and Advisors effectively
Implementing Structured Output and Prompt Engineering techniques
Designing AI-Powered Microservices using Spring Boot
Writing Integration Tests for MCP-based systems
Applying real-world AI architecture patterns and implementation best practices

By the end of the course, you will be able to:

Build production-grade AI Agents and Agentic Workflows using Spring AI and Java
Design and implement MCP Servers with Tools, Resources and Prompts
Integrate OpenAI, Gemini and local LLMs into Spring Boot applications
Build context-aware AI systems using ChatMemory, Advisors and Structured Output
Apply production-oriented AI architecture patterns, testing strategies and best practices

Throughout the course, we will build practical, production-style AI systems using Spring Boot, Spring AI and MCP.

Who this course is for:

Java and Spring Developers exploring AI Agents and MCP

AI Agents & Agentic Workflows with Spring AI, MCP and Java

What you'll learn

Explore related topics

Course content

Course Introduction & Roadmap1 lecture • 6min

LLM Fundamentals & Local Setup8 lectures • 30min

Spring AI: LLM Integration32 lectures • 1hr 53min

Spring AI: Context Management & Advisors12 lectures • 39min

Spring AI: Practice Test0

Building AI-Powered Microservices30 lectures • 2hr 7min

MCP Architecture & Core Concepts27 lectures • 1hr 21min

MCP Tool Design Patterns20 lectures • 1hr 1min

Agentic Workflows & Multi-Turn Reasoning9 lectures • 33min

Real-Time Progress Notifications7 lectures • 30min

Requirements

Description

Who this course is for: