🚀 My First Blog Post: WordPress + AWS + Containers + AI (What Could Go Wrong?)

So this is my first blog post. Strap in — this may get lengthy since I’m new to this and just starting.
My writing won’t be polished and fancy. I won’t send it through AI to get polished professionally and try to sound smarter than I am.I decided today I’ll just walk through the process I followed to actually create my WordPress site and deploy it to AWS.


🤔 Why WordPress?

First, let’s say I’m not a huge WordPress advocate. I’m simply jumping on the bandwagon since this is just a blog site.

  • ~42–44% of all websites use WordPress
  • ~60% of websites that use a CMS use WordPress
    As reported by W3Techs, the industry standard tracker

So I decided to take this simple concept and make it probably the most complicated I can by using AWS and containers.


🐳 Step 1: Local WordPress with Docker

First, I asked ChatGPT:

“Give me step-by-step instructions to run a local WordPress site using nothing but Docker containers.
One container for the site code and one container for the database.”

Without going into detail, it basically just walked me through creating a docker-compose.yml file with:

  • mariadb:10.6
  • wordpress:latest
docker-compose up -d

Wow — a running WordPress site immediately. I just browsed to:

http://localhost:8080

…and went through the setup steps in the WordPress installation web UI.


🌍 Step 2: Domain Setup (AWS)

Now I wanted to get a domain and register it. I used AWS for that, too.

Just bought the domain:

therobblog.com (you’re here)


🏗️ Step 3: “Let’s Make This Enterprise” 😅

For the rest of my setup, I decided to make this as enterprise-ready as possible… because millions of people are going to want to read my blog!

In AWS, I manually created:

  • A VPC
  • 4 subnets
    • 2 private (no direct internet access)
    • 2 public (for load balancers / ingress)
  • RDS MariaDB
  • Created database connect values as secrets in AWS secrets manager

🤖 Step 4: Let AI Drive (Mostly)

After some thinking (AI questions, Google searches, trying stuff myself, and using the AWS web console, which kind of sucks, etc.), I had a realization:

“Self – I’ve been using Claude Code and Codex for a few weeks now and have seen it everything fancy it can do for me in my code and run local commands right in my terminal, so why wouldn’t that same thing work for AWS since I have the awscli installed?”

So I basically told it:

“Hey, I want to deploy a wordpress container to AWS using ECS to get a wordpress image, configure it to use a MariaDB RDS DB I created and use Fargate serverless deployment for the container infrastructure”

And added:

“I want the wordpress container accessible on my public subnet and an ALB configured to send traffic to the fargate setup. I already have a MariaDB instance set up in AWS RDS with arn of <arn id here>. The connection parameters are stored in secrets manager with the ID of <arn id here>”. I I want all port 80 traffic to the ALB to forward to port 443 for SSL and then go to my wordpress site behind the load balancer.”

The AI went to work…..and when finished told me to go to the ALB DNS name in my browser.


🧪 Step 5: First Failure (Good Sign)

I hit the ALB DNS URL…

💥 Database connection error.

Honestly? Good sign.

  • Networking worked
  • ECS worked
  • ALB worked
  • WordPress container was running

🔧 Step 6: AI Fixes the Problem

After a bit more back-and-forth, the AI figured out:

👉 SSL issue with the DB connection

It fixed it and said:

“Try that URL again.”

💥 BAM — WordPress install page.

Pretty amazing stuff.


🌐 Step 7: Route 53 + DNS

Finally, I told it to configure Route 53 and DNS for my domain.

And again…

👉 It just did it.


⏱️ Total Time

This entire setup:

  • WordPress
  • Docker
  • AWS ECS + Fargate
  • RDS
  • ALB
  • Route 53
  • Secrets Manager

👉 Took about an hour


🧠 Final Thoughts on AI

I’m becoming more and more sold on AI.

The biggest realization:

The best way to use AI is to manage it like a person.

But also remember:

Like a person… it can be wrong.

  • Review everything
  • Understand what it’s doing
  • Don’t blindly trust it
  • Don’t underestimate what the AI can do for you.  It has come a LOOONG way.

You can also:

  • Use “plan mode”
  • Require approval before execution
  • Ask it to explain instead of act

✍️ That’s It (For Now)

Not polished. Not fancy. But real.

And honestly… kind of wild what’s possible now.

Leave a Reply

Your email address will not be published. Required fields are marked *

About Me

Rob Simmermon

Software Engineer

I’m a software engineer who’s equally at home in code and creativity. When I’m not building things, I’m into ATVs, acting, TV, and movies. This blog is where those worlds meet.

Follow Me

Connect with me and be part of my social media community.