AWS Builder Center
My First Steps into AI on AWS: What Amazon Bedrock Looks Like Through a Beginner's Eyes

My First Steps into AI on AWS: What Amazon Bedrock Looks Like Through a Beginner's Eyes

A first-timer's honest walkthrough of exploring AI on AWS — what Amazon Bedrock actually is, why "agentic AI" is suddenly everywhere, and what it feels like to try generative AI tools for the very first time without a machine learning background.

Three weeks ago, "AI on AWS" meant nothing more to me than a buzzword I kept seeing in job postings. I didn't know the difference between a foundation model and a fine-tuned one, and "agentic AI" sounded like something out of a sci-fi pitch deck. This post is about what it actually looked like to go from that starting point to opening Amazon Bedrock for the first time.

Why I started with Bedrock, not a tutorial

Most beginner advice says "learn the fundamentals first." I get the logic, but I learn better by poking at something real. Amazon Bedrock turned out to be a surprisingly low-pressure entry point — it's a managed service that gives you access to multiple foundation models (Amazon's own, plus others) without needing to train or host anything yourself. No GPUs to provision, no infrastructure to manage. You pick a model, send it a prompt, and see what comes back.
That mattered a lot for someone new. I didn't need to understand transformers or attention mechanisms to get a working result on day one. I needed to understand just enough to ask a good question.

The moment "agentic AI" clicked

A term I kept running into was agentic AI — AI systems that don't just answer a question but take multi-step actions to complete a task. At first this felt like marketing language. It clicked when I tried giving a model a goal instead of a single question — something like "help me draft and refine an email" instead of "write an email." Watching it reason through the task in steps, rather than just spitting out one response, was the first time AI on AWS felt less like a chatbot and more like a tool with judgment.
I'm nowhere near building actual AI agents yet. But I understand now why so many builders are excited about this shift — it's the difference between a tool that answers and a tool that works alongside you.

What surprised me most as a beginner

  1. You don't need an ML background to start. Bedrock is designed so application developers, not just data scientists, can build with foundation models.
  1. Cost visibility matters immediately. Just like with compute services, it's easy to lose track of usage if you're experimenting freely — which is part of why I'm also pushing for better built-in safety nets for beginners across AWS (see my Wishlist submission on Sandbox Mode).
  1. The AI space moves fast. Model options, pricing, and even terminology change often enough that a "beginner's guide" written six months ago can already feel dated.

Where I'm headed next

My next step is trying to build something small and real — probably a simple prompt-based tool rather than anything agentic yet. If you're also new to this, I'd genuinely like to hear what got you curious about AI on AWS, and what confused you the most when you started.
If nothing else, I hope this post is proof that you don't need to feel behind to start. I certainly did — and three weeks later, I've sent my first prompts to a foundation model and I'm still standing.
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