ESP32s, vibe coding, and the interesting threads around embedded software

My degree is in electrical engineering. I mostly chose it because it sounded cooler than “software engineer” at the time, and liked working with electronics as much as I liked software.

Arduinos were fun, but I was always time-poor and found the feedback loop slow. You’d carve out an evening, spend half of it re-learning the toolchain, another quarter debugging why the code is boot-looping, and maybe get an LED to blink before bed. One stalled evening and the project goes back in the box.

But the real difference between now and five or ten years ago is that LLM coding tools exist. I’m comfortable working through an AI, and that changes the shape of what’s viable when you’re time-poor.

Last month I re-picked up a NeoPixel clock I’d stalled on for five years. The hardware had survived one kid and one house move. I gave the project to an AI and it one-shotted the whole thing in ten minutes. The missing link turned out to be state management in C++ classes. Something I’d never wrapped my head around solo.

That unblocked something. I started poking around and found projects like RSVPNano and the Waveshare ESP devices: prototyping platforms with everything enclosed. Instead of needing to be mailed a whole suite of parts and waiting for the next delivery to continue, you just unbox and go. That shift alone makes a difference when your tinkering time comes in unpredictable thirty-minute chunks.

I saw a twitter post about one of those little ESP32 S3 devices that are suddenly everywhere. A Digital fidget toy to replace scrolling: Opus 5 wrote the renderer and sim in C to get the most out of it. That’s the whole appeal in a single sentence: you pick up one of these prototype boxes, tell the LLM what you’ve got plugged in, and it finds the right pins and flashes it. No wiring, no waiting for delivery, no digging through forum threads to remember which GPIO does what. A 2.06in AMOLED ESP32 Waveshare device shows up, you plug it in, and you’re making things happen on it within minutes.

The iteration loop shrinks from “an evening” to “twenty minutes while the kids finish dinner.” You spend your mental energy on what you want to build, not on remembering which website had the working BLE example.

There seems to be a little renaissance happening in tech circles around these devices. It’s really easy to ask a frontier model to write the firmware and have an okay-good time. And honestly, I’m just enjoying the prototype boxes at the moment. You don’t need to wire anything. The hardware problem is solved by the form factor itself.

I’ve also noticed my five-year-old is getting to the age where writing a prompt and seeing something happen in the real world is genuinely interesting to them. That’s a whole other angle I hadn’t considered when I blew the dust off that clock.

I have a few things in mind, mostly around an ed-tech physical device. So keep following for more.

For anyone who’s wanted to play with hardware but never had the margin: this is the moment. You don’t need to hold the full platform in your head anymore. You need to know what you want to build, and be able to describe it clearly. The rest is fast, cheap, and the LEDs light up on the first try more often than they have any right to.