AI has been all the rage as of late, and I’ve been using it for various tasks. From code completions to scaffolding complete web apps. Before I can really get into a project using one language model or another, it feels like there’s a new model on the scene that promises to be better than the last.

So, I’ve been learning my cadence with these models, trying to understand their strengths and weaknesses without getting too caught up in the hype cycle.

In reality, it’s often best to understand the underlying capabilities of LLMs in general, then apply that knowledge to whatever model is currently available. After all, these models are all based on similar architectures and training methodologies, so understanding one gives you a leg up on the others.

What am I using for my projects? That depends.

For coding, I will combo up with GitHub Copilot and OpenAI Codex. With Copilot, I have access to various models, from OpenAI ChatGPT to Anthropic Claude. Each has their strengths and weaknesses, so I pick and choose based on the task at hand.

For example, if I have small, incremental changes to make to some codebase, then perhaps GPT-5 is the right model…however, if I want to go brrr on code, then Claude 4 is my go-to for generating large chunks of code quickly.

For writing blog posts, (I’m on VS Code) Copilot attempts to help me write, however, it’s distracting. So, I try to limit how much AI is injected into my flow by ignoring most of the suggestions.

Oh… and OpenAI’s Sora… that’s been fun to play with… see my video below for a demo of Sora in action…

My buddy Oscar and I hanging out, drinking a beer, while waiting for some pizza. We took a selfie and when the Ghibli craze hit, I used our image to have some fun. Well, now with Sora, I was able to animate the image and have the two characters interact. Watch below.

I don’t know, maybe start a podcast with pure AI co-hosts? Hmmm… 🤔… Or, maybe not and just leave animation and podcasting to the professionals?

Check out my profile on Sora and hit the follow button, jsldvrsora.

A couple of other projects I’m toying with are jsldvr/fishing-report and the jsldvr/react-template-static-rss-reader which both leverage AI in different ways to enhance user experience.

The Fishing Report project was entirely vibe coded using a combination of GitHub Copilot and OpenAI Codex. I describe it as, “A web app that provides science-based fishing predictions for North America. Combines lunar phase calculations, real-time weather data, and optional almanac ratings to generate bite scores for optimal fishing times.”

The Latest News web app, uses the static rss reader project to return both articles and YouTube videos. It pulls from various tech news sources and presents them in a clean, easy-to-read format. This project was also built with the help of AI, making it a breeze to set up and customize.

Both projects are open source and are linked on the nav of this website.Anyway, aside from enjoying the blessings of God above and liberty, that’s what I’ve been cooking up lately.