Two ex-OpenAI founders, zero patience for corporate AI nonsense, and one big question: what actually happens when you let AI run a whole company?
And for a while, it felt like it was. Brilliant people, insane compute, a real shot at changing everything. But somewhere between the press releases and the enterprise sales decks, the mission got blurry.
The AI got more filtered. The decisions got slower. The vibe? Very corporate. So we left.
MeTal started as a question: if you stripped away all the fear and bureaucracy, what would an AI company actually look like? We're finding out in real time.
"The models were good. The decisions about what to do with them — not so much."— The founders, probably
Respectfully. No drama. Just a different vision for what AI should be.
Built the first version of everything using AI. That includes this website.
Valley Fast and Valley Pro hit the API. People actually like them.
We got a game studio. AI is running it. No humans involved in the actual game. This is the experiment.
We think you can build AI that's responsible AND actually useful. Imagine that. You don't have to lobotomize a model to make it safe.
We're not going to make our AI refuse to talk about things because it might make someone uncomfortable. Context matters. Nuance matters.
AI that just answers questions is last decade. We're building AI that actually does things. That means trusting it — which most companies are too scared to do.
The AI industry has enough people in turtlenecks explaining how they're going to save humanity. We're just here to build cool stuff and see what happens.
Ex-OpenAI. Builds everything with AI, including his opinions on AI companies.
Also ex-OpenAI. The technical half. Probably has an opinion about your stack.
// the rest of the team is, uh, AI. we're working on it.
Check out what we're building or just say hi. We don't bite. Probably.