Optional analytics

With your permission, analytics help us understand page visits, clicks, form progress, engagement, and technical errors. On selected public pages, Microsoft Clarity also provides a masked session replay. We mask all text and form entries. These tools are optional, and you can withdraw permission at any time. See our privacy policy.

← Back to blog
For Recruiters

What is OpenAI Dots? ChatGPT's always-on AI agent explained

Sep 30, 2026 · 6 min read

HOW A DOT WORKSGive it a goal, and it keeps going1You set a goalTell it what you wantand what it may dowithout asking you.2It works aloneOn its own cloudcomputer, with theapps you connected.3It asks firstSensitive steps waitfor you. Passwords andmoney stay with you.4You reviewIt brings back thefinished work for youto check and approve.Between your messages, it keeps reading your connected apps for ways to help.

On 29 September 2026, OpenAI launched dots at its DevDay event in San Francisco. OpenAI describes them as "remarkably capable, always-on agents built to handle everything". A dot is an AI agent that lives inside ChatGPT. It keeps working on the goals you give it after you close the chat.

OpenAI writes the name in lower case, and it calls each one "your dot". Everyone starts with one dot. You give it a name, and you can pick a character or a pet as its picture. The rollout began on launch day, and OpenAI says it may take several days to reach every account.

What is a dot?

In a normal ChatGPT chat, nothing happens until you type your next message. A dot keeps track of what you asked for and carries on with the work between your conversations. When it needs a decision from you, it sends you a message.

OpenAI gives each dot the following:

  • Its own computer in OpenAI's cloud, with its own web browser. You can open that computer at any time and watch what the dot is doing.
  • Connections to your apps. Through ChatGPT's plugins, a dot can connect to more than 4,000 apps, e.g., your email.
  • A memory. A dot receives memories from ChatGPT and keeps its own notes, so it learns your preferences over time.

The AI model behind every dot is GPT-6 Astra. OpenAI calls Astra its most capable model and its best model for using a computer.

What can a dot do?

One of OpenAI's examples comes from an early tester. He had forgotten to send an invoice to a publication. His dot noticed the missing invoice and prepared it. It sent the invoice after he approved it.

Inside OpenAI, staff use dots for project work. When someone reports a bug in Slack, dots start investigating it straight away. OpenAI's launch post also describes a dot that reads customer feedback and builds small fixes. It then brings the finished code to a developer to review.

You can also give a dot regular jobs. You can ask it to set a reminder or run a recurring check, e.g., a review of your calendar every morning.

When you are not working with it, a dot looks for ways to help on its own. OpenAI calls this proactive research. The dot reads the apps you have connected and makes notes. OpenAI says the tools it uses for this can only read, so they cannot send messages or change anything.

Where can you talk to your dot?

You create a dot in the ChatGPT desktop app or in ChatGPT in a desktop web browser. You cannot create one on a phone. Once it is set up, you can message it in the ChatGPT mobile app.

A dot also works in Slack and Microsoft Teams. It keeps the same context wherever you reach it, so you can start a project in ChatGPT and follow up in Slack. You can also call it to talk through a problem. Texting is in a small beta for Pro users in the United States. At launch, a dot cannot call you.

What does a dot cost?

Dots come with ChatGPT Pro, OpenAI's top personal plan, and with Business Premium for companies. Pro now comes in three monthly plans:

  • Pro 100 costs $100 a month.
  • Pro 200 costs $200 a month.
  • Pro 500 costs $500 a month. It is the only Pro plan with Ultrafast, OpenAI's fastest speed setting.

Your first dot is included at no extra cost. Talking to your dot does not count against your ChatGPT limits. Larger jobs that it starts in OpenAI's other tools, e.g., Codex for coding, count against your plan as usual. OpenAI says you will later be able to pay for more dots, or for a dot that works faster or takes on more.

PLANS AND PRICESWho can get a dot at launchPLANPRICEDOTSChatGPT Pro 100$100 a monthFirst dot includedChatGPT Pro 200$200 a monthFirst dot includedChatGPT Pro 500$500 a monthFirst dot included, plus UltrafastBusiness PremiumCompany planIncluded in every supported regionEnterprise, Edu, HealthcareCompany planBeta, off until an admin turns it onFree, Go and PlusCheaper plansNot includedPro users in the EEA, Switzerland and the UK are left out at launch. Dots are not open to anyone under 18.
Who can get a dot at launch, from OpenAI's help pages.

Who can get one?

OpenAI's help page says Pro users can get dots in every market except the European Economic Area, Switzerland and the UK. Business Premium users can get them in every region where ChatGPT is offered. India and the UAE are not on the excluded list, so Pro subscribers there should see dots as the rollout reaches their accounts. Meta's Muse, by contrast, is still open only in the United States and Canada.

Enterprise, Edu and Healthcare workspaces can try dots as a beta. A workspace admin has to switch it on first. Dots are not available to anyone under 18.

What stays in your control?

Every dot starts with OpenAI's rules about when it may act alone and when it must ask. You can add your own, which OpenAI calls Custom Rules. For each kind of action, you pick one of four settings:

  • Take action without asking.
  • Take action if pre-approved. This means you asked for that exact action in your request.
  • Ask before taking action.
  • Hand off to you. You finish the step yourself.

Some limits stay fixed whatever you choose. The most sensitive actions always come back to you. OpenAI names changing a password and moving money.

A separate OpenAI system called auto-review also checks some actions before they run. Before your dot sends an email, auto-review checks who it is going to and what it says. It is there to catch problems like a wrong address.

You can follow a dot's work in the Activity View of the desktop app, and you can pause it at any time. The memory controls are limited at launch. You cannot see or delete a single memory. Disconnecting an app also does not remove what the dot has already learned from it. The only way to clear a dot's memory is to reset the dot, which deletes it.

What could a dot do for a hiring team?

Recruiters spend a lot of time on follow-up work, e.g., chasing feedback after interviews. OpenAI built dots for ongoing work of this kind. A recruiter with a dot could ask it to:

  • Remind interviewers to send their feedback the day after each interview.
  • Put the job description, the candidate's CV and past notes into one short brief before each call.
  • Keep a shortlist up to date as new candidates apply.

Candidate data needs care. When you connect a dot to your inbox or to an applicant tracking system, you give it access to other people's personal details. OpenAI says it does not use data from Business and Enterprise workspaces to train its models by default. On personal plans, a setting controls this. Check your company's rules before you connect anything that holds candidate records.

A dot can help a candidate prepare for an interview. The interview itself is where the employer hears the candidate answer in their own words.

Where this leaves dots

Dots are one day old, and many accounts do not have them yet. For now, you need to pay at least $100 a month for ChatGPT, or work at a company on Business Premium, to use one. OpenAI says it plans to bring dots to more users soon.

Frequently asked questions

Related posts

Run your first round with AI interviews

Screen resumes, interview every candidate by web or phone, and decide from evidence. Book a 30-minute walkthrough on one of your real roles.

Book a demo