[Speaking Engagement Report] I presented on the topic "Getting Started with AI Agent Utilization Through Amazon Bedrock AgentCore"

Introduction

Hello, I'm Jinno from the Consulting Department, and I love the supermarket La Mu!

Recently, I gave a presentation at a customer seminar on the theme "Getting Started with AI Agents Using Amazon Bedrock AgentCore" (the official seminar title included a catchphrase, but I'm only including the core part in this article).

The content covers the basic concepts of AI agents and a brief explanation of Amazon Bedrock AgentCore. I thought this information would be useful to everyone, so I decided to share the materials publicly!

Presentation Materials

Content Explained in the Seminar

In this seminar, I mainly explained the following topics.
Here's a brief introduction to the content.

Basic Concepts of AI Agents

First, I carefully explained "What is an AI agent?" starting from the basics. I did this because technology updates related to generative AI have been happening very rapidly lately, and I thought many people might be confused despite hearing these terms frequently.

AI Agent Demo Introduction

I introduced how AI agents work using simple use cases and demonstrations. I'm currently preparing to make the application shown in the demo available to the public after some further refinements. I hope you're looking forward to it! (Of course, it's created based on AgentCore)

Image of Agent-based RAG

I believe it's important to incorporate the concept of "Human in the Loop" where human approval or action is involved for critical decisions, rather than automating everything with AI agents.

Determining how much authority to delegate to AI agents is a key design consideration. How much work we allow AI agents to do should be carefully considered during the design phase.

Utilizing Amazon Bedrock AgentCore

Up to this point, we've been explaining AI agents, but when implementing them on AWS, Amazon Bedrock AgentCore stands out as a strong option.

From here, I provided specific explanations of Amazon Bedrock AgentCore's features, focusing mainly on the Runtime. I emphasized the importance of combining necessary functions (Identity, Memory, Gateway, Built-in Tools, Observability) based on the type of AI agent you want to create.

For more details on the features, please refer to the presentation materials or the blog post below where I've summarized the key points.

https://dev.classmethod.jp/articles/amazon-bedrock-agentcore-developersio-2025-osaka/

Conclusion

In this presentation, I focused on explaining the conceptual aspect of "What is an AI agent?" in depth. Building on that, I briefly introduced how to implement this specifically on AWS and how to create AI agents using Amazon Bedrock AgentCore.

I hope that through the actual demonstrations, those who attended this presentation could better visualize how AI agents work, and I plan to actively share samples of AI agents for various use cases in the future!

I would be delighted if this presentation and materials made you think "AI agents are interesting!" or "I want to try Amazon Bedrock AgentCore!"

Thank you very much for reading until the end!!