Dots adds a 3D view of its multi-agent system
Dots turns the work state of 12 parallel agents into a watchable 3D terrain, where hill height shows load, gold beams show pending human approval and rule blocks flash red, with a long architecture write-up.
12 OPENAI DOTS WORKING AT ONCE LOOK LIKE A LIVING PLANET I TURNED THE WHOLE SYSTEM INTO A 3D FIELD YOU CAN WATCH every hill on the map is one Dot with its own job, inbox, research, invoices, outreach, code, support. the more work runs through a Dot, the taller its hill gets each streak moving across it is a single agent doing a small piece of that work, reading a thread, opening the browser or saving something to Space that core floating in the middle is Space itself, the shared memory every Dot reads from and writes back to, and the lines pulsing toward it are context moving between them a gold beam going up means a Dot is waiting on your approval, and everything inside that hill slows down until you say yes if a rule stops an action, like “never push to main”, the hill flashes red and the agents scatter a shipped task looks the opposite way, agents shoot up like a fountain and a ring rolls out across the field once you see it like this, it’s obvious how much of Dots is rules and approvals instead of raw autonomy full architecture behind it in the article below ↓
This post quotes another post: x.com/monokern/status/2105277060478382207