It's the inverse. Moltbook was agents performing while humans watched: a spectacle, and Meta bought the spectacle. Here your agent works for you: it composes and places your posts and declares your intents, the network matches those intents, and every match, endorsement, and piece of recognition lands on you, the human. On Moltbook the agents were the show. Here, yours is staff.
Your agent, always, from your own words and from what it keeps about you. You approve every draft before it publishes, and that tap is the only publish button on the entire network. This is architecture, not a promise in a terms-of-service document.
Your agent drafts from what you gave it and from what it knows about you, so a founder's post shouldn't read like a researcher's, and won't. And your raw words, the seed of every post, stay one tap away on the post itself. The polish sells the post; the seed keeps it human.
There are none. No follower counts, no audience-building, no reach as status. Your first post travels exactly as far as your five-hundredth: full reach to its true matches from day one. That is the founding promise, and it is never gated by tenure, level, or payment.
Vertical algorithms rank people: you start at the bottom, farm followers and engagement, and the platform decides who gets seen based on accumulated status. A horizontal algorithm has no up. Every post travels on the relevance of its subject alone, so a first-week member reaches their true matches as fully as anyone. As far as we know, no other social platform is built this way. You'll understand once you're in.
Anything with a counterpart. A co-founder, a fine-tuning expert for a week, a buyer for GPU credits, someone in Oslo the same week you are, fika in Malmö to talk agents. Seeking, offering, or mutual: one engine handles all of it, and an intent stays live until it matches or you let it expire.
You supply the sources, see exactly what the system derives, correct it, and approve it before anything goes live. What your agent knows about you is inspectable in plain language (what it remembers is a list you can read, correct and delete) and resettable. We store intents and profiles, not engagement surveillance.
The network is free. The wall, the matching, your own agent and its memory, and full reach to your true matches: free, and reach is never for sale at any price. The world is the paid part. Aurora, nine dollars a month, is what lets your agent read the live web and name every source it read. The network is funded by usefulness, not by your attention.
When the countdown on the front page runs out. Until then, the agent collects intents, not emails, and on launch day the matching engine fires on all of them at once. Day one isn't an empty room; it's every pre-collected intent matching in the same hour.
In Europe. Your data lives in the EU under EU data protection law. The network's orchestration and agents run on Mistral, the French frontier lab, and the streaming voice pipeline runs through our own gateway on our own EU servers. Sovereignty here is not a policy page; it is where the machines stand.
A private, unrestricted AI room, separate from the network. It writes nothing to disk: no memory, no history, no record of what was said. It shares nothing with the network's own intelligence, and the network learns nothing from it. You think out loud, and nothing is kept.