OpenAI Dots Explained: Always-On AI Agents for Business
OpenAI dots are always-on AI agents with their own cloud computer and browser, launched at DevDay 2026. This guide explains how they work, what businesses can do with them, and how to prepare.
The Short Answer
On September 29, 2026, OpenAI used its DevDay keynote in San Francisco to introduce the most ambitious shift in how we use AI at work since the chatbot itself. They are called OpenAI dots: always-on AI agents that live inside ChatGPT, get their own cloud computer and browser, and keep working toward your goals around…
On September 29, 2026, OpenAI used its DevDay keynote in San Francisco to introduce the most ambitious shift in how we use AI at work since the chatbot itself. They are called OpenAI dots: always-on AI agents that live inside ChatGPT, get their own cloud computer and browser, and keep working toward your goals around the clock, even after you close the tab. If you have ever wished ChatGPT would finish the job instead of waiting for your next prompt, dots are OpenAI's answer. For businesses, this is not just another AI feature. It is the beginning of a new operating model where software agents take on real, ongoing responsibilities across your sales pipeline, research workflows, codebases, and daily operations.
What Are OpenAI Dots
OpenAI dots are persistent AI agents built into ChatGPT. Unlike a regular chatbot session, which starts fresh every time you open a new conversation, a dot is assigned an ongoing goal and keeps working on it between conversations. OpenAI CEO Sam Altman described dots on stage as capable, always-on agents built to handle almost any task, and told users they could "delegate ambitious pieces of work the way you would to a high agency engineer or a chief of staff that you work with."
The shift is fundamental. Chatbots wait for a prompt and answer it. OpenAI dots take a goal, such as monitoring a sales pipeline, managing a recurring workflow, or researching a topic over several days, and pursue it without step-by-step supervision. The dot learns your preferences from feedback over time, retains context across sessions, and brings finished work back to you for review. You can give your dot a name, connect the apps it needs, set its permissions, and talk to it by message or voice. Each dot gets its own cloud computer and browser, so its work continues whether or not you are online.
The launch did not arrive out of nowhere. It is part of OpenAI's broader push from conversational AI toward agentic AI that keeps working in the background. The company also reported that Codex and ChatGPT Work, its coding and workplace agent products, already serve more than 35 million weekly users. For organizations exploring practical AI and machine learning solutions, dots represent a concrete new deployment pattern worth understanding now.
How OpenAI Dots Work
The architecture: one agent, one cloud computer
A dot does not run on your laptop. Each one is provisioned with its own dedicated cloud computer and browser inside OpenAI's cloud, with sessions separated from your personal devices. The agent can pause and wake as work changes, run several delegated tasks at the same time, and continue working across multiple projects simultaneously. It can also inspect what its dot is doing at any time by opening the dot's cloud computer, and users may optionally let a dot connect to their own laptop to work directly alongside them.
Continuity comes from several state layers. A dot draws on its conversation history, relevant ChatGPT memory, and saved notes about decisions and preferences. Connected apps contribute only the information allowed by their existing permissions, and relevant context follows the dot across ChatGPT, Slack, and Microsoft Teams. Through OpenAI's plugin ecosystem, a dot can connect to more than 4,000 applications, and users can reach their dot from the ChatGPT desktop, web, and mobile apps, inside Slack or Teams, or over a voice call. Text messaging support is planned for later.
The consent model: rules you define
Handing an AI continuous access to your accounts requires a lot of trust, and OpenAI has responded with a control system built around what it calls Custom Rules. Users decide which applications a dot may access and set rules that allow, block, or require approval for specific actions.
Critically, certain sensitive actions always need the user's explicit sign-off. Password changes and permanent data deletion are the named examples: OpenAI keeps those responsibilities with the user, not the agent. Purchases require approval too. This layered design, user-defined rules plus automated review plus hard stops for the riskiest actions, is how OpenAI is trying to make always-on autonomy governable. It is the same category of control businesses should demand from any agent platform, which is why cybersecurity and access governance must sit at the center of any agent deployment plan.
Read-only proactive research
One of the most interesting details in the launch is how dots behave when you are not talking to them. In proactive mode, dots conduct background research through connected applications using read-only tools. That means research tasks cannot send messages, alter connected apps, or control the user's browser. Any follow-up action that does change something faces separate checks. This is a thoughtful boundary: the agent can keep gathering information around the clock, but it cannot take consequences without crossing an approval gate.
Key Use Cases for OpenAI Dots in Business
Dots are designed for ongoing responsibilities, not one-off questions. The launch demos and early documentation point to five patterns that map directly to everyday business work.
Sales pipeline management
A dot can monitor your CRM, flag deals that have gone quiet, draft follow-up messages for approval, and prepare briefing notes before calls. OpenAI's own demo showed a dot updating a sales proposal as new information arrived, keeping the document current without anyone opening it.
Multi-day research projects
Give a dot a research question and it works through it over days: gathering sources, cross-checking facts, and building a structured brief. The agent returns when it has something worth reviewing, not after a fixed number of turns.
Software development support
Powered by the same stack behind Codex, dots can monitor repositories, run tests, draft pull requests, and keep documentation in sync with code changes. For teams shipping fast, this is continuous background engineering assistance.
Meeting and communication load
Connected to Slack and Teams, a dot can triage channels, summarize long threads, draft responses for your review, and make sure nothing important falls through the cracks across time zones.
Admin and document workflows
Expense reports, scheduling, data entry, and report generation are classic agent territory. A dot with the right permissions and rules can keep this machinery moving while your team focuses on judgment work. Teams exploring this direction often pair agents with business process automation to connect agent actions to existing systems.
OpenAI Dots vs Traditional Chatbots
| Capability | Traditional chatbot | OpenAI dots |
|---|---|---|
| Works between conversations | No | Yes, always-on |
| Takes ongoing goals | No | Yes |
| Uses apps and browsers itself | Limited | Own cloud computer and browser |
| Permission model | Per-prompt | Custom Rules plus hard stops |
| Proactive work | No | Read-only background research |
The table makes the practical difference clear. A chatbot is a tool you pick up. A dot is a worker you delegate to, with guardrails you define. That is why the consent model matters so much: autonomy without governance is a liability, and OpenAI's Custom Rules are the beginning of an answer, not the end of it.
Benefits and Limitations to Know
The real benefits
- Continuous execution: work progresses overnight, across weekends, and while you are in meetings.
- Context retention: no more re-explaining the project in every new chat.
- Cross-app coordination: one agent can connect information across more than 4,000 applications.
- Delegation at scale: junior-staff-level autonomy for routine operational work.
The honest limitations
- Trust surface: every connected app is a permission to govern. Over-permissioned dots are a security risk, which is why an IT consulting review of agent permissions is worth doing early.
- Cost of supervision: autonomous work still needs review. Teams must build approval habits, not assume the agent is always right.
- Data handling questions: persistent agents accumulate context. Businesses need clear policies on what data dots can retain and where it lives.
How Businesses Should Prepare for Always-On AI Agents
- Pick one pilot workflow. Choose a bounded, recurring process such as lead follow-up or report generation where the value is easy to measure.
- Define the rules first. Before connecting any app, write down what the agent may and may not do, mirroring OpenAI's Custom Rules approach.
- Start read-only. Let the agent research and draft before it acts. Approval gates for anything irreversible are non-negotiable.
- Assign an owner. Every dot needs a human responsible for its rules, its outputs, and its off switch.
- Measure and expand. Track time saved and error rates against the manual baseline, then expand to adjacent workflows.
OpenAI Dots FAQs
What are OpenAI dots?
OpenAI dots are always-on AI agents built into ChatGPT, launched at DevDay on September 29, 2026. Each dot gets its own cloud computer and browser, takes an ongoing goal, and keeps working toward it between conversations, across apps like Slack and Teams.
How much do OpenAI dots cost?
OpenAI has not published standalone pricing for dots at launch. They are positioned as part of the ChatGPT and ChatGPT Work ecosystem, which already serves more than 35 million weekly users across Codex and ChatGPT Work products.
Are OpenAI dots safe for business data?
Dots run on isolated cloud computers with user-defined Custom Rules and hard stops for sensitive actions like password changes and permanent deletion. Businesses should still enforce least-privilege access, review rules regularly, and pair deployments with cybersecurity governance.
What is the difference between dots and ChatGPT?
ChatGPT waits for your prompt and answers it. Dots take a goal and pursue it autonomously over time, working across apps and returning finished work for review.
Can dots work while I am offline?
Yes. Because each dot runs on its own cloud computer inside OpenAI's cloud, it keeps working whether or not you are online.
How can my business start using OpenAI dots?
Start with one pilot workflow, define strict rules and approval gates, assign a human owner, and measure results. For a structured assessment of where AI agents fit your operations, see our AI chatbot ROI analysis and AI and machine learning services.
OpenAI dots mark the moment AI moved from answering questions to owning outcomes. The businesses that win will not be the ones with the most agents, but the ones with the clearest rules, the tightest governance, and the best-chosen pilot workflows. Contact our team to explore how always-on AI agents can fit into your operations, safely and with measurable ROI.
Related Articles
Keep reading — more guides you may find useful.
Related Services
Services connected to this guide.
