Hack Your Eyewear: Building Smart Glasses for Under $25

You've probably seen smart glasses with price tags in the hundreds or even thousands of dollars. They promise augmented reality, AI assistants, and hands-free computing—but they also demand a serious investment. What if you could build your own pair for less than the cost of a pizza dinner for two? That's exactly the premise of OpenGlass, an open-source project that turns any ordinary pair of glasses into a hackable smart device using off-the-shelf components you can source yourself.

What It Does

OpenGlass is exactly what it sounds like: open-source smart glasses. The project provides everything you need to transform a regular pair of glasses into a wearable device that can record video, recognize people you meet, identify objects, and translate text in real time. It's powered by a Seeed Studio XIAO ESP32 S3 Sense board, a small microcontroller with built-in camera and microphone capabilities, paired with a compact 250mAh battery.

The architecture is split into two main parts. On the hardware side, you've got the ESP32 board and battery housed in a 3D-printed case that mounts to your glasses. On the software side, there's a firmware component (an Arduino .ino file you flash to the board) and a companion application built with Expo, a React Native framework. The app handles the AI heavy lifting—it connects to services like Groq and OpenAI for vision and language processing, or you can run everything locally with Ollama if you prefer to keep your data off the cloud.

One important note: the project maintainers have moved active development to a new repository called Omi. The original repo still contains all the documentation and code you need to build one, but future updates will happen over there.

Why It's Cool

There's something genuinely refreshing about a project that treats smart glasses as a weekend DIY project rather than a luxury purchase. Here's what makes OpenGlass stand out:

• The price point. Under $25 for the core components is remarkable. You're not paying for custom silicon or proprietary hardware—you're using a $15-ish dev board and a $10 battery. The magic is in the software.

• It's genuinely hackable. Because everything is open source, you can modify the firmware, change the AI backends, or build entirely new features. The ESP32 platform has a massive ecosystem, so you're not locked into anyone's walled garden.

• Multiple AI backends. You have choices here. Use Groq for fast cloud inference, OpenAI for their vision models, or run Ollama locally for complete privacy. That flexibility is rare in consumer hardware.

• The form factor. A 3D-printed clip-on case means you can attach this to any glasses you already own. No need to buy a specific frame or deal with proprietary designs.

• The community angle. It's backed by Based Hardware, and there's an active Discord for getting help and contributing. This isn't an abandoned weekend project—it's a real community effort.

The cleverest design decision is probably the separation of concerns. The glasses capture video and audio, but all the processing happens on your phone or computer via the Expo app. That keeps the hardware cheap and power-efficient while still giving you access to state-of-the-art AI models.

How to Try It

If you want to build your own pair, head over to the OpenGlass repository and gather your components. You'll need:

• A Seeed Studio XIAO ESP32 S3 Sense board
• An EEMB LP502030 3.7v 250mAH battery
• A 3D printer (or access to one) for the glasses mount case, with the STL file provided in the repo

Once you have the hardware, the setup process looks like this:

1. Print the mount case using the provided STL file.
2. Flash the firmware by opening the .ino file in the Arduino IDE. You'll need to add the ESP32 board package and select the XIAO_ESP32S3 board. Don't forget to set PSRAM to "OPI PSRAM" in the Tools dropdown before flashing.
3. Set up the software by cloning the repo and installing dependencies:

git clone https://github.com/BasedHardware/openglass.git
cd openglass
npm install

1. Add your API keys in the keys.ts file for Groq and OpenAI. If you're using Ollama, self-host the REST API and set the URL to http://localhost:11434/api/chat, then pull the vision model with ollama pull moondream:1.8b-v2-fp16.
2. Start the app with npm start (or yarn start if you prefer). It's an Expo project, so you'll open the localhost link to access the web version.

The setup is a bit involved—you'll need to be comfortable with Arduino IDE, basic command-line work, and API keys. But the README walks you through every step, and the firmware readme offers an alternative arduino-cli path if you're more comfortable with terminal-based builds.

Final Thoughts

OpenGlass is a great example of what's possible when you combine cheap commodity hardware with modern AI APIs. It's not going to replace a polished consumer product—the setup requires some tinkering, the battery life on a 250mAh cell won't last all day, and you'll need to be comfortable with flashing firmware and managing API keys. But that's kind of the point. This is a project for people who want to experiment, who want to understand how their tech works, and who might have ideas that go beyond what commercial smart glasses offer.

Given that development has moved to the Omi repository, you might want to check there for the latest updates. But the original OpenGlass repo remains a complete, functional starting point for your own smart glasses adventure. If you've ever wanted to wear your code on your face, this is a pretty affordable way to start.