OpenAI dots are always-on AI agents inside ChatGPT that run on their own cloud computer, connect to over 4,000 apps and keep working toward a user's goals between conversations. For marketing teams, a dot can act as a teammate that researches, drafts and updates work in ChatGPT, Slack and Teams, with approval rules on anything sensitive. It is also a new kind of researcher reading about every brand, so being easy for a dot to find, read and verify is now part of the job.

OpenAI introduced dots on September 29, 2026, announcing them in its DevDay 2026 recap and describing them as "remarkably capable, always-on agents built to handle everything." Most AI assistants have waited for a prompt and stopped when the conversation ended. Dots change that model. A dot starts with context from the user's ChatGPT memory, looks for ways to help in the background, and messages the user when a decision needs them. That matters for marketers in two directions. Inside the team, a dot can carry recurring work such as reporting updates and content drafts. Outside the team, a buyer's dot can research vendors continuously, long before the buyer opens a browser tab. This guide covers how dots work, how they compare with ChatGPT Pulse, who can use them in 2026, and what marketing teams should do about them.

What OpenAI Dots Are and How They Work

OpenAI's launch post and its dots product page describe five defining traits.

1. A dot is always on. OpenAI says dots "can work towards your goals 24/7," and a dot messages its user with progress, questions or decisions that need them.

2. Each dot has its own cloud computer. Dots run on their own cloud computer with their own browser, separate from the user's laptop unless the user chooses to connect it.

3. Background research is read-only. When the user is not working with it, a dot does proactive research using tools "restricted to be read-only," drawing on permitted connected sources. According to OpenAI, those research tools cannot send messages to other people, change content in connected apps, or control a browser or desktop.

4. It connects to over 4,000 apps. Through plugins, OpenAI says dots "can readily connect to over 4,000 apps." Users choose which apps a dot can access and what permissions it holds.

5. It lives where teams already talk. Dots are reachable in ChatGPT on web, mobile and desktop, and can be brought into Slack channels and Microsoft Teams. Texting a dot is a separate limited beta for US Pro users and is not available in Business or Enterprise workspaces. Dots are powered by GPT-6 Astra.

How Dots Handle Approvals and Safety

OpenAI's safety, security and privacy post for dots sets out the controls that decide what a dot can do alone.

• Custom Rules. Users can allow specific actions, require approval for them or block them, such as telling a dot never to send emails. Dots can help write Custom Rules, but changing them needs the user's approval, and rules cannot remove mandatory confirmations or core safety requirements.

• Auto-review. A separate safety system checks planned steps, such as sending emails or changing files, against the user's instructions, Custom Rules and safety requirements. A blocked step does not run.

• Steps that always stay with the user. Changing a password and moving money between financial accounts are handed back to the person. Purchases with cards saved on merchant sites require approval.

• Monitoring and credentials. OpenAI monitors dots as they plan and carry out actions, and if safety monitoring flags a concern, a dot can pause that work and show a warning. Saved passwords are provided through an encrypted credential service "without passing it to the model."

OpenAI also notes that dots "can still make mistakes," so consequential work should be reviewed before it ships.

OpenAI Dots vs ChatGPT Pulse and Agent Mode

Dots arrive after two earlier OpenAI attempts at proactive and agentic work, and OpenAI has retired both.

• ChatGPT Pulse ran asynchronous research overnight and delivered it the next morning as visual summaries, and was offered to Pro users. OpenAI's Pulse help article states that "Pulse is being sunset as proactive updates move into scheduled tasks," and the ChatGPT release notes dated the announcement June 17, 2026, with Pro access continuing for 14 days. Pulse researched and suggested. A dot researches continuously, acts on connected apps within its rules, and works in Slack and Teams.

• ChatGPT agent mode carried out tasks started from a conversation, using a virtual browser. OpenAI's ChatGPT agent help article now states that "ChatGPT agent is no longer available" and points users to ChatGPT Work for multi-step tasks and to cloud browser for browser workflows.

The practical difference is persistence. Pulse delivered a daily briefing and agent mode handled a single task, while a dot holds goals over time and comes back with finished work.

Availability and Pricing in 2026

Access depends on plan and region, and the rules matter for European teams. OpenAI's getting started article for dots sets them out, and notes that availability is subject to gradual rollout.

• Pro: Available to Pro users "in markets excluding the European Economic Area, Switzerland, and the UK."

• Business Premium: Available "across all supported ChatGPT regions," which makes it the route for teams based in the EU, Switzerland or the UK.

• Enterprise, Edu and Healthcare: A beta that a workspace admin must enable, and it is "initially turned off by default."

• Pricing and limits: OpenAI says the first dot is included in Pro or Business Premium "at no extra cost." Conversations with a dot do not count toward ChatGPT usage limits, while tasks started in Codex or ChatGPT Work do.

• Data and age: OpenAI's dots privacy, security and safety FAQ states that content from Business, Enterprise and Edu workspaces is not used for training by default, and that dots are not yet available to users under 18.

Using a Dot as a Marketing Teammate

OpenAI's own examples map closely to marketing work. In one, a content creator's dot learns their voice, finds clip moments in interview transcripts, and drafts show notes and social posts for approval. In another, a product lead's dot revises launch materials and prepares drafts of asset and doc changes when the product scope changes. On the dots page, a dot updates a launch FAQ in Slack and drafts a manager update in Teams.

Four workflows suit a marketing team in 2026:

• Market and competitor monitoring. A dot can be tasked with tracking competitors, category news and pricing pages through its browser, while its background research keeps reading connected sources such as inboxes and shared docs.

• Recurring reporting. A dot with read access to analytics and ad accounts can refresh weekly reports and flag changes that need a decision.

• Content repurposing. Long-form assets become drafts of posts, newsletters and show notes, held for approval before anything is published.

• Launch coordination. A dot in the team's Slack channel can answer recurring questions, keep FAQs current and pick up next steps.

RZLT recommends setting Custom Rules before connecting any publishing or email tools: require approval for anything external, and block sending outright until the team trusts the dot's output.

How to Be Findable When a Buyer's Dot Does the Research

A buyer's dot can browse the web with its own browser when it is given a research task, so the brands it recommends are the ones it can read, connect to and verify.

1. Make pages extractable for agents. A dot reading a site needs clear product descriptions, visible pricing and structured data it can parse without guesswork. RZLT's guide to how AI browsers change the way buyers find brands covers the page-level changes, from structured data to extractable answers.

2. Ship a plugin. Dots reach apps through plugins, and OpenAI's developer docs state that "Apps are now submitted and published as plugins." Plugin submission requires a verified publisher identity, a public MCP server and review before listing, and the plugin is published in the countries the developer selects.

3. Keep claims consistent and verifiable. A dot that researches over days can cross-check a brand's site against reviews, comparison pages and third-party coverage. Inconsistent pricing or claims across those sources weaken the case for recommending the brand.

4. Cover the paid layer in ChatGPT. Organic visibility and ads work together inside ChatGPT. RZLT's definitive guide to ChatGPT Ads for advertisers covers formats, targeting and EU availability for teams adding paid reach.

How Marketing Teams Should Prepare for OpenAI Dots

Dots move OpenAI from answering questions to holding goals, which changes both how marketing teams work and how buyers research them. RZLT recommends a simple order for 2026:

• Pick the right plan first. Teams in the EU, Switzerland or the UK need Business Premium or Enterprise, since Pro excludes those markets.

• Start with read-only work. Point the first dot at monitoring and reporting, review its output for a few weeks, then add drafting and channel work.

• Audit how a dot sees the brand. Ask a dot to research the brand and its competitors, then fix the gaps it reports in pages, pricing and proof.

• Decide on a plugin. Products with a natural task inside ChatGPT should plan an MCP-based plugin and its review timeline.

The underlying shift sits in RZLT's B2A Manifesto on marketing to personal AI agents: when agents become the gatekeepers, marketing has to meet their criteria of facts, fit, price and proof. RZLT works with OpenAI's tools directly, and partner Josip Vlah is the official OpenAI Ambassador for Croatia, a role built around workshops and builder sessions on AI agents in Croatia and its neighboring AI communities.