A task can continue after the conversation

OpenAI introduced dots on 29 September: persistent agents inside ChatGPT that work toward ongoing goals using GPT-6 Astra, a cloud computer and connected apps. The useful change is continuity. A dot can carry a project forward between conversations, rather than waiting for a fresh instruction at every step.

OpenAI describes work such as turning customer feedback into proposed software fixes, adapting launch materials when a product changes, and preparing clips, show notes and social copy from interview transcripts. These are vendor examples, not results measured by XANTRANS. The announcement also previews specialist dots for organizations, initially through focused enterprise pilots.

What a small team could hand over

Consider a studio preparing a launch. A useful first assignment would be to follow an approved project folder, compare new feedback with the brief, and assemble a revised copy deck with a list of unresolved decisions. The team would receive something concrete to review, instead of another summary of what still needs doing. This is a suggested application, not a workflow we have tested.

OpenAI's getting-started guide says dots can also start tasks in Codex cloud environments that you have already created. Local computer access is optional and starts disabled. That distinction matters: an agent's cloud workspace does not automatically give it access to everything on your laptop.

The Baltic availability catch

At launch, personal Pro access excludes the European Economic Area, Switzerland and the UK. That excludes Latvia, Lithuania and Estonia from the initial Pro rollout. Business Premium is listed across supported ChatGPT regions; Enterprise beta access requires an administrator to enable it. Rollout is gradual, so an eligible plan does not guarantee immediate access.

Initial setup happens in the desktop app or desktop web. Check the official getting-started page for your account before changing plans: launch access and usage terms can change.

Ongoing work still needs boundaries

The privacy FAQ distinguishes proactive research from taking action. Its research tools can read permitted sources and retain notes, but cannot directly send messages, modify connected content or control a browser. Follow-up actions remain subject to instructions and safety checks. Custom Rules can add restrictions, such as preventing email sending.

There is also a memory limitation worth understanding before connecting company systems. Disconnecting an app stops new access but does not erase context already retained. OpenAI says individual dot memories cannot currently be edited or deleted separately; deleting the dot resets its own context. Start with a narrow source set and a clearly defined output.

Sources & verification

Prepared with AI assistance from the linked official sources. XANTRANS has not independently tested dots; suggested uses are illustrative.