Announced 29 Sept 2026 · Sources checked
How is ongoing work different from a prompt?
A single request usually has a clear end. An ongoing responsibility needs a trigger, an expected result, and a stopping rule.
| Define | Example |
|---|---|
| Trigger | A model provider publishes a release |
| Output | A sourced draft for editorial review |
| Allowed action | Gather sources and prepare the draft |
| Completion | Editor receives a draft with supporting links |
Where should a team start?
Start with one inspectable output, such as a sourced brief. Use the first results to refine the instructions and source list before expanding the job.
What makes the feedback loop useful?
A reviewer should be able to see what changed and why.
- Include the source supporting each change.
- Label work as proposed or already completed.
- Track review time as well as accepted output.
Common questions
Are dots available to every account?
No. The announcement describes plan and market limits. Check OpenAI’s current availability information.
Has AiLookout tested dots?
No. This article analyzes the concept in OpenAI’s announcement.
What to remember
An always-on agent needs a well-defined responsibility, a useful feedback loop, and a clear boundary around its access.
Sources & further reading
How this story was made
Written by Kristian Kostov with AI assistance and checked against the linked sources. Company performance claims are attributed to the company. Analysis reflects AiLookout’s interpretation; we have not independently tested the products discussed. Cover photography is illustrative and does not depict the specific announcement or product.
Our editorial standards
