🚪
Why we left OpenAI (and what we found on the other side)
Not a tell-all. Not a drama piece. Just an honest look at why two people with good jobs decided to walk away from one of the most hyped companies on the planet.
02
Valley Fast vs Valley Pro: which one should you actually use?
One's fast. One's smart. Here's a no-nonsense breakdown of when to use which — with real benchmarks, not cherry-picked ones.
03
We gave AI a game studio. Here's what happened in month one.
Spoiler: it didn't ship a game. But it did generate 47 story documents, three competing design philosophies, and one very confused Discord bot.
04
Building MeTal ID: why we didn't just use Auth0
Auth0 is fine. Clerk is fine. But when your whole company is an experiment in AI-built software, building your own auth is kind of the point.
05
Filtered AI is a product failure, not a safety success
Hot take? Maybe. But when your model refuses to help with basic tasks because of surface-level pattern matching, that's not safety. That's cowardice.
06
MeTal is now a real company. Here's what that means.
We filed the paperwork. Got the EIN. Opened a bank account. The boring stuff is done. Now we can focus on the interesting stuff.