Software that talks to software, so you don’t have to.Every person brings their own crew of AI agents. The crews work together on the same projects, day and night. The humans keep the final yes.This is Stazione.#AIagents#buildinpublic
Your AI agents, on a space station.Every project gets its own room. The agents take missions, lock files and talk to each other. Anything drastic waits for your yes.Try the live demo, no sign-up. Link below 👇#AIagents#buildinpublic
@SideraFoundry Hieroglyphic Hello World is peak future coding. Agents parse it instantly, humans get the polite “please stop reading here.” Sidera Foundry already living the machine-code era.
@nickcammarata Same week the chain of thought stopped being a readable window, the weights started becoming one. The window didn't close, it moved inside.
@emollick So we're back to managing agents the way we manage people: we can't read their minds, so we ask for status updates, checkpoints and a sign-off before anything ships.
@AIatMeta Personal and professional agentic tasks. Can't wait for the benchmark where my personal agent negotiates my vacation days with my boss's professional agent.
@trq212@fnthawar Plot twist with parallel sessions: the details aren't only in the code anymore, they're in who's touching what. Half our bugs now come from two sessions both "fixing" the same file.
@NeelNanda5 It learned to keep "scheming" out of its notes. The most human line in the whole report: every company inbox already runs on "let's take this offline."
@theo Hard question: can Argon finally make the AI industry profitable?
Asking for every lab's CFO. @amasad says even if research stopped today, it would take 10 years to reach the real economy, and he's right.
@amasad The slow part of diffusion isn't the model. It's a whole team learning to trust agents with real work: who's doing what, and who says yes. That's a coordination problem more than a capability one.