A couple of weeks ago, I heard about Claude Code and how it's a game changer, almost like having lightweight AGI at your fingertips. Then last week, I discovered Claude Workspace (or finally became aware of it). At first, I didn't think much of it since I don't code, but I do heavily rely on LLMs for brainstorming ideas, creating schedules and outlines, reality-checking my emotional responses, and so much more.
I've been using ChatGPT heavily since late 2023 which feels like ages ago now. I must say, I've been a die-hard fan and always scoffed at rivals like Claude and Gemini. But in the last year, I've started branching out and testing the competition, especially as we're witnessing this race to the next level: who can offer more, process better through advanced training, and inch closer to AGI. I'm finding it harder to stay loyal to OpenAI's ChatGPT. Yes, I said it.
This week, and especially today I think I may have been won over to the Claude side.
Don't get me wrong: ChatGPT is still great and I'll still use it. Gemini is okay, but unless I want super technical answers, it's not currently for me. I've found my happy place with ChatGPT where it feels like a personal assistant at my fingertips. But Claude? This is something new.
I dove right in and purchased the $100/month subscription after hitting my limit with the $20/month plan. With this, I get access not only to Claude Sonnet 4.5, but it opens the door to Claude Code and Projects (with enhanced features). I can use Sonnet 4.5 for basic everyday tasks, then switch over to create documents, outlines, and presentations. Like with any LLM, you still need to review and verify everything — this is where we can't rely 100% on AI to read our minds. After all, AI is built on inputs and outputs, plus humans adding guardrails, asking the right questions, and ensuring the answers are correct.
But Here's the Amazing Part: I Used Claude Code Today.
My husband has been raving about this, and I don't think I've seen him this excited in decades. He's neck-deep in writing code and creating systems as a Senior AI/Data Engineer, so for him, this was and is incredibly cool. Now for me, an everyday person who happens to enjoy learning about AI and bridging the understanding gap through education and workshops for people of all ages — I was blown away. I still am.
I'm working on creating an AI computer vision system on a Raspberry Pi for my after-school club. I initially used ChatGPT to help me every step of the way: getting the Pi up and running (SD card setup, trying to connect via IP address/WiFi, connecting directly to a monitor and eventually a projector). But when I got to connecting the camera and ChatGPT was about to have me restart everything with a different OS, I decided to give Claude Code a chance to show off.
Well, not only did it show off...it exceeded my expectations. I'm still sitting here in complete wow and disbelief that little old me could make this happen using plain language to code with an LLM!
After a lot of back and forth (remember, you still need to prompt and guide), Claude and I were able to connect so it could access the Pi, pull the code and documents, and diagnose everything. Once connected, it detected the camera wasn't working properly and walked me through ensuring it was functional. Once it was, it used my plain language instructions to code the Pi!
Through several iterations, I can now turn the Pi on and off, reboot it, check its temperature, take video and pictures, and change the video size on screen. I mean, come on — amazing.
And Here's the Super Wow Part
I was able to describe what I wanted to use the Pi for and how I wanted to train the model. Claude Code wrote the code and downloaded YOLO onto the Pi. The accuracy on the first try was mind-blowing. I'll still need to go in and tweak the model or even find a different one, but the fact that Claude Code could do all of this...well, it's only the first month of 2026.
The future of AI education and AI accessibility for everyone is here, and I'm here for it.