AWS Builder Center

Builder Center

Series Overview: Agentic Applications with Amazon Bedrock

Series Overview: Agentic Applications with Amazon Bedrock

Introduction and table of contents for the series on building agents with Strands Agents.

Jason Stehle

Series contents

Once you are done with this series, check out the Amazon Bedrock AgentCore Starter Toolkit .

Series intro

Welcome to what is shaping up to be a very long series about building agentic applications. This initial batch of articles covers agentic fundamentals and intermediate topics using Strands Agents.
The learning curve for developing generative AI applications can be very steep. The rate of change is extremely high, and the amount of hype is distracting. We'll try to focus on practical use cases as much as possible. We might do something impractical for learning purposes, or maybe just for fun.
We'll be using the Strands Agents SDK  and Python to learn the fundamentals and build our demos. Strands Agents has a great combination of ease of use and power. But more importantly, it gives you the ability to understand what's happening under the hood when you need to. This will be a recurring theme throughout the series.
By the end of the series, hopefully you:
  • Have a solid understanding of agentic concepts and techniques
  • Can build moderately sophisticated agentic proofs of concept
  • Know how to build your prototypes so they can eventually be moved to production without too much rework

Assumptions about you

We're going to assume the following:
  1. You're familiar with working at a command line.
  2. You're comfortable working with code.
  3. You have a code editor installed, like Visual Studio Code 
  4. You have API access to AWS and Amazon Bedrock, either through your employer or through your own account.
 

A note on the examples

Let's get one thing out of the way now: many of the examples in this series are not true "agents" (depending on your definition). It's totally fair criticism. However, it's not our goal to pass a purity test. Our goal is to help builders understand and use the major components of agentic applications.
Many of our examples are barely more sophisticated than the prompt/response generative AI of past years. But these examples should help you better understand how to apply agentic techniques. The most effective learning examples aren't always going to be purely agentic. This has the benefit of reflecting the next generation of systems with generative, agentic, and deterministic components.
It's also important to know that these examples are meant to be demo applications. They lack polish, error handling, observability, and rigorous testing. They are nothing more than a starting point towards validating use cases.
 

Set up a Python 3.13 virtual environment

Python 3.13 is recommended for these exercises. Other versions 3.10+ should probably also work, but have not been thoroughly tested.
You can download the Python 3.13 installer from here: https://www.python.org/downloads/ 
Once Python is installed, we can set up our project.
First, create a new directory for our code, and cd into it.
Next, create a virtual environment to keep our dependencies isolated:
1
python3.13 -m venv .venv
On MacOS or Linux, run the following command to activate the Python environment:
1
source .venv/bin/activate
On Windows, run:
1
.venv\Scripts\activate
 

AWS and Amazon Bedrock access

To set up access to AWS & Amazon Bedrock, please see the Amazon Bedrock documentation. 
To set up Python with AWS, please see the AWS Boto3 library documentation. 

Special thanks

Thank you to the following folks for reviewing & providing feedback on this content:
  • Jyothi Madanlal
  • Tanner McRae
  • Justin Muller
  • Magdalena Nedelcu
  • Kris Schultz
  • Fahim Surani
  • Narcisse Zekpa
# generative-ai# strands-agents# agentic-ai# amazon-bedrock
Any opinions in this article are those of the individual author and may not reflect the opinions of AWS.
Enjoyed reading this content? Let the author know!

Your likes, comments, shares, and saves help creators reach more builders.

Loading recommendations

Loading article