WML X.X.0

WML X.X.0 is a privacy-first desktop pet that reacts to your activity, battery, security and workflow, offering helpful local AI suggestions while staying fun, customizable, and non-intrusive.

Comment

Inspiration

WML X.X.0 was inspired by the idea that a desktop assistant does not have to feel cold, invasive, or purely functional. I wanted to build something that could live on your computer like a small companion: present, useful, and playful, but still respectful of your privacy.

I also have a website called whitemirrorlab.com, where I share experiments with the world, both social experiments and personal ones. WML X.X.0 fits into that same spirit: it is a personal experiment about how software can feel more alive and human without becoming intrusive.

What it does

WML X.X.0 is a floating desktop pet that reacts to your local activity, battery level, and workflow signals. It earns points as you use your computer, lets you customize the pet with animals, colors, and accessories, and can suggest practical actions through local AI.

It also includes an optional OpenAI-powered assistant. If the API is not configured or fails, the app falls back to local logic, so the core experience still works offline.

How I built it

I built it as a desktop app using Electron, React, and TypeScript. Electron handles the native desktop window, transparent always-on-top behavior, local system signals, and packaging. React powers the interface, settings panel, shop, and assistant chat.

The app uses local state persistence, activity signals, battery data, process detection, and a small local rule-based AI layer to decide when suggestions should appear. OpenAI integration is optional and only receives summarized, anonymized state.

Challenges I ran into

One of the biggest challenges was balancing usefulness with calmness. A proactive assistant can easily become annoying, so I had to design limits around when suggestions appear and what actions are allowed.

Another challenge was connecting native desktop behavior with a polished frontend: transparent windows, draggable UI, activity detection, battery changes, animations, and multiplatform builds all had their own quirks.

Privacy was also a major design constraint. I wanted the app to react to user behavior without reading personal content or sending unnecessary data away.

Accomplishments that I'm proud of

I am proud that WML X.X.0 feels more like a companion than a traditional productivity tool. It has personality, visual customization, local intelligence, and useful actions, while still staying lightweight and privacy-minded.

I am also proud of the fallback design: the app can use an OpenAI model when available, but it does not depend on it. The local AI layer keeps the experience functional and practical by default.

What I learned

I learned how much work goes into making desktop software feel natural. Small details like window behavior, suggestion timing, visual feedback, and state persistence matter a lot.

I also learned that AI does not always need to be the center of the experience. Sometimes the best AI feature is one that quietly supports the product, respects boundaries, and only appears when it can genuinely help.

What's next for WML X.X.0

Next, I want to improve the pet's personality, add more animations, expand customization, and build a safe web demo so people can try the experience without installing anything.

I also want to keep improving the local AI layer, add more useful actions, and make WML X.X.0 part of a wider set of experiments shared through whitemirrorlab.com.

Built With

Share this project:

Updates