> ## Content Index > Fetch the complete content index at: https://nextwith.ai/llms.txt > Use this file to discover other available public pages before exploring further. # OpenAI’s Dots bring persistent AI agents into ChatGPT, and business controls become the test - URL: https://nextwith.ai/openais-dots-bring-persistent-ai-agents-into-chatgpt-and-business-controls-become-the-test/ - Published: 2026-09-30T07:43:17.000Z - Updated: 2026-09-30T07:43:17.000Z - Description: OpenAI is rolling out Dots as persistent ChatGPT agents for eligible paid users. The company describes read-only background research and action approvals; business buyers should verify those controls in a supervised pilot. - Author: NextWith.ai Editorial Desk - Tags: AI Agents, News OpenAI used its DevDay stage on September 29 to introduce Dots, a new agentic assistant that the company says can keep working in the background and pursue user goals with limited oversight. That matters because the product is not just another chat feature. If OpenAI’s description holds, Dots pushes ChatGPT closer to delegated software: something a user can set in motion, leave running, and expect to act across tools and time. According to [TechCrunch’s report on the announcement](https://techcrunch.com/2026/09/29/openai-launches-dots-its-bubbly-agentic-avatar?ref=nextwith.ai), OpenAI says Dots are meant to work independently of a specific device or interface. The company’s framing is that a user can name a primary Dot, assign goals, and eventually have multiple Dots cooperate on that person’s behalf. OpenAI says [Dots are rolling out](https://openai.com/index/introducing-dots/?ref=nextwith.ai) to Pro and Business Premium users in eligible markets, while Enterprise users can try the beta when an administrator enables it. TechCrunch reports that users can launch Dots from Codex or ChatGPT. That rollout detail is important for readers trying to understand who should care first. This is not a universal consumer feature at launch. The staged rollout and administrator-controlled Enterprise beta make Dots a workflow product as much as a model showcase. For companies, that means the real question is not whether an agent can chat convincingly. It is whether an always-on agent can be made useful without creating a new layer of unattended authority inside the business. The mechanism OpenAI is describing goes beyond prompt-and-response. TechCrunch reports that users can message Dots through Slack, Teams and other workplace platforms, with text-message support coming later. OpenAI also says specialist Dots can be given identities, credentials and tools through existing systems. The company’s own examples include a developer dot monitoring customer feedback and a scientist dot rerunning analysis when new data arrives. Those are useful illustrations because they show the intended pattern: Dots are meant to observe, decide and act over time, not merely draft text on demand. That design creates the central trade-off. The more useful an agent becomes, the more access it needs. If a Dot can read internal messages, use credentials and take actions in connected systems, then a deployment depends on permissions, logging, revocation and human escalation, not just model quality. In practical terms, a business would need to decide which actions a Dot may suggest, which it may execute automatically, and which should always wait for approval. OpenAI has described some of those controls, but its product announcement is not an independent demonstration of how they work in a company’s own systems. OpenAI says a Dot works on a separate cloud computer unless the user elects to connect a local machine. It says background “proactive research” uses read-only connected tools, while action review, custom rules and approvals govern more consequential steps. These are company descriptions, not independently tested guarantees. TechCrunch reports that OpenAI is also working with Microsoft to integrate specialist Dots into Agent 365 security controls; OpenAI describes those as focused enterprise pilots. Identity, permissions and auditability are therefore the practical tests before deployment beyond a limited pilot. A useful test would begin with a low-risk connected app, inspect what the Dot can see during background research, and then try an action that requires approval. The company says users can inspect a Dot’s cloud computer, but that visibility still needs to be evaluated in the actual workspace configuration. Dots also arrives in a broader race over autonomous agents. [The Guardian reported](https://www.theguardian.com/technology/2026/sep/29/openai-announces-dots-agent-safety-concerns?ref=nextwith.ai) that the launch follows Meta’s Muse by weeks and comes amid heightened scrutiny of agent safety. That context does not establish a flaw in Dots. It does make the distinction between an announced safeguard and a demonstrated one more consequential for organizations deciding what an agent may do on their behalf. Dots may be a significant step toward persistent workplace agents, but a buyer should test the controls OpenAI describes before granting access to internal systems: confirm app permissions, review the activity record, and try the approval path on a consequential action. The product announcement establishes the intended design; a supervised pilot must establish how it behaves in practice.