Udemy
    •  
    •  
    •  
    •  
    •  
    •  
    •  
    •  
Subscribe
Turn what you know into an opportunity and reach millions around the world.
Learn More
Your cart is empty.
Keep shopping
AI Agent Memory Architecture with Spring AI
Hot & New
New
Rating: 5.0 out of 5(7 ratings)
140 students

AI Agent Memory Architecture with Spring AI

Build layered memory systems with Java, Spring AI, PostgreSQL, pgvector, and scalable backend architecture
Hot & New
New
Created byInfiproton Tech, Harish B N
Last updated 5/2026
English

What you'll learn

  • Design AI agents with layered memory using Spring AI Advisors and PostgreSQL
  • Build persona, episodic, semantic, and working memory for AI assistants
  • Implement vector-based memory retrieval using pgvector and embeddings
  • Create AI systems that remember users correctly across conversations
  • Build scalable async memory pipelines for production-style AI backends
  • Develop backend AI applications that learn user preferences over time
  • Understand why chat history alone is not real memory for AI agents

Course content

7 sections • 27 lectures • 1h 49m total length

  • What “Memory” Really Means in AI Systems2:19

    Understand what memory means in AI systems and why LLMs do not inherently remember users.

  • Why Chat History Is NOT Real Memory3:00

    Learn why replaying chat history is not enough for building memory-aware AI agents.

  • Memory Architecture Overview (L1–L4)3:19

    Explore the four logical memory layers used in modern AI assistants.

  • Travel Planner Use Case2:17

    Introduce the AI travel planner project used throughout the course.

  • Project Setup + Minimal Chat7:38

    Set up Spring Boot, Spring AI, PostgreSQL, and build a minimal AI chat endpoint.

  • Source Code & Git Branch Guide1:11

Requirements

  • Basic knowledge of Java and Spring Boot is recommended
  • Prior experience with SQL databases like PostgreSQL will help
  • No prior AI or machine learning experience is required
  • Curiosity about how modern AI assistants remember users across conversations

Description

Most AI applications do not truly remember users.
They simply replay chat history.

In this course, you will learn how to design and implement real memory systems for AI agents using Java, Spring AI, PostgreSQL, and pgvector.

Using a practical AI Travel Planner project, you will build a layered memory architecture that enables AI assistants to remember users correctly across conversations.

This is a backend engineering focused course designed for developers who want to move beyond basic chat applications and build production-style AI systems.

What You’ll Build

  • Working memory using conversation history

  • Persona memory for persistent user facts

  • Episodic memory using conversation summaries

  • Semantic memory using learned preferences

  • Vector similarity search with pgvector

  • Async memory processing pipelines

  • Centralized prompt assembly using Spring AI Advisors

What You’ll Learn

  • Why chat history is not real AI memory

  • How modern AI memory systems are structured

  • How to design layered memory architectures

  • How embeddings and vector search work in practice

  • How to retrieve relevant memory dynamically

  • How to build scalable AI backend pipelines

  • How to personalize AI behavior across conversations

Technologies Used

  • Java

  • Spring Boot

  • Spring AI

  • PostgreSQL

  • pgvector


By the end of this course, you will have a complete understanding of how real AI memory systems are designed and implemented in modern backend applications.

Who this course is for:

  • Backend engineers building AI assistants and copilots
  • Spring Boot developers exploring AI memory systems
  • Java developers interested in production-style AI architecture

Report abuse