Field noteSep 30, 20262 min read1 source

Openai dots agent feature claims

In one minute

OpenAI’s dots announcement is a move from conversational assistance toward delegated, persistent work.

Thesis

dots frames ChatGPT as a supervised agent that carries project work forward between conversations.

Observed facts

  • The official page says dots use ChatGPT memory, Codex, and connected tools, with users choosing which apps they can access.

    1
  • The page says dots run on their own cloud computers and that some actions affecting accounts or sharing information require approval.

    1
  • The page presents dots as powered by GPT-6 Astra and says they are created in the ChatGPT desktop app.

    1

Mechanism

dots run on their own cloud computers, while user-managed app permissions and Custom Rules constrain what they can do; actions that could affect accounts or share information pass through Auto-review, and some require approval.

Why now

OpenAI is presenting persistence, tool access, and bounded autonomy together as a product, rather than describing an agent solely as a chat interface.

Implications

  1. teams evaluating agentic workflows should assess the permission and approval model alongside task capability, because action scope is part of the product design.

Evidence boundary

the source is OpenAI’s product page, which describes intended capabilities and illustrative workflows, not independent performance results or verified deployment outcomes.

Watch conditions

  1. look for concrete availability details and a documented list of supported integrations before treating broad connector claims as established.

Evidence / sources

  1. 1chatgpt.com/features/dots
All Field Notes

Next reads

Read next

Open the graph
  1. 01archiveField NotesOpen the latest editorial analysis.
  2. 02source ledgerSource LedgerInspect the public source evidence graph.
Who read this page?Machine requests, hidden until opened

Loading the privacy-safe route aggregate...

Open the JSON contract