In partnership with
OpenAI introduced Dots on September 29 as agents that can continue work between conversations. The announcement marks a shift from a single prompt and response toward ongoing tasks. Dots can use connected apps and run on a separate cloud computer, according to OpenAI. The company says people can inspect the computer's work and set rules for what an agent may do. These are product descriptions, not independent proof of reliability.
7 Stocks to Ride The A.I. Megaboom
The next A.I. boom could create massive winners just like the 1990s tech surge.
We identified 7 small tech companies positioned to benefit from the next phase of A.I. growth.
See them inside this free report 7 Stocks to Ride The A.I. Megaboom.
The design joins several familiar agent functions. A user gives a goal, connects relevant services, and defines permitted actions. OpenAI says Dots can keep working between chats, return results for review, and operate through ChatGPT, Slack, or Teams. The release also describes a browser and cloud computer for each Dot. This setup separates the agent's computing environment from a user's device unless access is granted. It does not remove the need to consider connected accounts and stored information.
The important change is duration. A chatbot usually acts after a user sends a request. A persistent agent can retain context and resume tasks later. That may reduce repeated setup, but it makes memory and access controls part of routine operation. OpenAI says users choose apps and specify rules; sensitive actions may need approval, while money transfers require user takeover.
The privacy FAQ describes a specific control limitation. Users cannot view, edit, or delete individual memories held by a Dot. Deleting the Dot removes its own context, but files and conversations it created remain stored separately. Disconnecting an app stops new access through that connection; it does not remove information already added to the Dot's context. This distinction matters when a task draws from several sources over time.
Memory sharing adds another layer. OpenAI says a Dot can receive ChatGPT memories and recent conversation context. Conversations with a Dot can also contribute to ChatGPT memory. Turning off ChatGPT Memory stops further sharing, but does not erase information the Dot already received. OpenAI also says human review can occur in limited safety-related circumstances, even when model improvement is disabled. These are the company's own descriptions of its controls and data practices.
Access is rolling out gradually rather than reaching all accounts at once. OpenAI says Dots are being introduced across Pro, Business Premium, and Enterprise plans in eligible markets. It says access will expand, but the announcement gives no date. Availability can change, so this is a dated account rather than a permanent access map.
Build or buy your support agent? Both skip the number.
Buying or building your customer service agent both carry a cost the pitch skips: maintenance if you build, flexibility if you buy. Running Agents in Customer Work is four conversations on agentic AI in customer ops, this one on build vs. buy. Register now, four Tuesdays, 10 a.m. PT.
Independent coverage confirmed the launch at OpenAI's September developer event. The Associated Press also reported that the company had delayed a different advanced model one day earlier because of safety concerns raised by researchers. That was a separate model decision, not evidence of a Dot failure. Keeping the events distinct matters: the launch establishes a product capability, while the delay illustrates scrutiny around the wider field.
The inspected sources contain no independent evidence about Dots completing long-running tasks across real connected accounts. Nor do product materials establish that permission rules prevent every unintended action. Key questions are whether users can understand retained context, keep permissions narrow, and audit results before consequential steps.
Persistent agents make memory governance part of everyday software design. Their utility and risk depend on duration, connected services, permissions, and correction options. Dots introduce a specific version of that design, with controls described by its maker and gaps documented in its own help material. This check-in records the current release boundary, not a final assessment of agent reliability.
The agentic era needs a different CRM. That’s Attio.
Teams like Parallel, Turbopuffer, and Wordsmith are already setting the pace on Attio. Get an always-on revenue engine, with agents and workflows that build pipeline, chase every buying signal, and move deals forward with your team. Whether you're working in your browser, inbox, or favorite agent, connect to your customer data in real-time through Attio's web app, MCP, API, and SDK.
