Sep 30, 2026 · 24 min
OpenAI pushes always-on agents into everyday life
Dots are OpenAI's cute lil guys
Dots show how AI companies are moving from chatbots toward persistent systems that can act across personal services, while raising harder questions about trust, cost, and control.
- 1Dots extend ChatGPT into continuously running agents that communicate across channels and complete tasks through connected services.
- 2OpenAI’s adoption challenge is balancing a more involved setup against the easier integrations offered by Google and Meta.
- 3Privacy failures, usage limits, and high computing costs could determine whether personable agents become practical consumer products.
Don't miss
The discussion of a Meta Muse failure involving a Facebook Marketplace seller’s address crystallizes the privacy risks of background agents.
The brief
David Imel and Hayden Field examine OpenAI’s Dots, continuously running agents designed to connect personal accounts, communicate across channels, and handle tasks beyond ordinary ChatGPT chats.
The central promise depends on integrations: Dots become useful when they can work across connected services, but OpenAI’s more involved setup may limit consumer adoption against Google and Meta.
Safety remains unresolved. The discussion revisits Meta’s Muse agent and a Facebook Marketplace address incident as evidence of why background agents must be deliberate before gaining broader powers.
OpenAI also faces practical constraints: one Dot per account, substantial computing demands, and uncertain free access complicate the idea of an always-available personal assistant.
The episode closes by placing Dots in a wider race involving Anthropic and Google, alongside protests at OpenAI’s Dev Day and the industry’s long-running bet on agent businesses.