OpenAI Dots Persistent Assistants

Diving deeper into

OpenAI

Company Report
Dots, announced in September 2026, operates its own cloud computer and connects to more than 4,000 apps through plugins, competing with Meta Muse and Instinct for an ongoing assistant relationship.
Analyzed 5 sources

This is OpenAI moving from chat sessions to becoming the default operating layer for a user’s digital life. Dots are not just another assistant UI, they are persistent workers that keep their own cloud computer running, stay connected to thousands of apps, and continue handling inbox, calendar, travel, purchasing, and follow ups after the user closes the chat. That makes the product less like search and more like an always on executive assistant with transaction upside.1 2 3 4 5

  • The product architecture matters. A dot runs on its own cloud computer and browser, so tasks do not depend on a user leaving a laptop awake or a browser tab open. That turns agent use from one off prompts into recurring background work, the same shift that made Codex usage scale beyond human keyboard time.1 2 3
  • The competitive set is defined by distribution and app reach, not just model quality. Meta is pushing Muse through WhatsApp, Instagram, Facebook, Messenger, and Threads, while Instinct is built around connected apps and devices for task execution. OpenAI’s answer is plugin breadth, persistent memory, and direct ownership of the execution environment.1 3 4 5
  • The business prize is commerce, not only subscriptions. Instinct is reported to be nearing $1B in annual transaction volume, with more than half tied to travel. That shows how a personal agent can sit in the middle of booking, buying, canceling, and rebooking, then monetize with take rates, referral fees, or merchant demand gen.4 5

The next step is a land grab for the trusted agent relationship. As users assign more dots, raise workload limits, and route more purchases through them, the winner will be the assistant that is easiest to trust with real permissions, best at finishing tasks across apps, and most deeply embedded in everyday communication surfaces.1 2 3 4 5