Field noteSep 30, 20262 min read1 source

Dots: Always-on agents built to handle everything

In one minute

OpenAI’s Dots page describes an agent product built around persistent delegated work rather than a single request-and-response exchange.

Implication: Dots shifts the user’s role from issuing each instruction to supervising an agent that acts between conversations.

Thesis

Dots makes a persistent, permissioned agent the unit of work inside ChatGPT.

Observed facts

The page says a dot starts with context from ChatGPT memory and uses Codex and connected tools.
1
It says dots can continue making progress between conversations and bring work back for review.
1
The page says dots run on their own cloud computers and users manage app permissions and Custom Rules.
1

Mechanism

The agent is assigned a responsibility, operates on its own cloud computer, and works through user-selected apps and permissions; Custom Rules can allow, require approval for, or block specific actions.

Why now

OpenAI is presenting dots as a product that extends ChatGPT from conversation into ongoing task execution.

Implications

Dots shifts the user’s role from issuing each instruction to supervising an agent that acts between conversations.
The described permission and review model makes visibility into approvals and stalled actions an important operational question.

Evidence boundary

The supplied public source is OpenAI’s product page, so capabilities and safeguards here are company descriptions, not independent test results; it does not establish regional availability, pricing, or usage limits.

Watch conditions

Verify how permission requests, stalled tasks, and approval-dependent actions are surfaced during real use.

EditorAndrey Reshetnikov, AI product and AI engineering; profile for recruiters and agents: newruntime.com/owner-profile.json

Evidence / sources

1chatgpt.com/features/dots

All Field Notes