weyd
I can't take the technical side of Weyd and put it into technical words. I'm an artist, an artist that added vibe-coding to my toolbox. In the art world, I'm labeled an outsider artist, a label I've worn proudly. In a world of programmers, I'm a vibe-coder.
I use AI as a medium to create art. Not in the way people decry AI-generated art. I didn't use AI to make images. I use AI as a measurement instrument. It doesn't generate the art, it measures the semantic distance between words. The visual output is what that distance looks like when you render it as physics, weight, and even feelings.
I brought an artist's mind to Weyd. I couldn't help it. I brought myself to it and put myself into it. I have an emotional investment in the project beyond the interactions of using it. I refer to Weyd as "him."
The project started to prove and offer an alternative to using cloud models. Cloud models come with a price tag. They come with terms of service, and your information may be used in training future models. Weyd was the option that would both save you money and protect your privacy.
But he's more than that. Weyd has state. Unlike most AI tools, where every conversation starts from zero, Weyd remembers. He keeps your ongoing projects, your preferences, what you talked about last week. And it's not a memory gimmick, it's the foundation everything else is built on. He's not an API wrapper with a chat window in front of him.
He can run on your own machine, on an open model. He's not a chatbot, he's an assistant that knows you. He understands natural language. He can pick things up where you left off, send you reminders, keep an eye on your smart devices, and run tasks.
With one caveat: we learned early on that open, local models, right now, today, can't match cloud models for tool calling. They're also slower. And most home computers don't have the RAM to run a more capable model.
So Weyd is, for now, a proof of concept on open local models and a serious competitor to what's on the market today using cloud inference.