What OpenAI Dots are, and the open-source way to run that work
OpenAI Dots are always-on agents that keep working between conversations, on a cloud computer OpenAI runs and under rules kept in ChatGPT's settings. Kortix is the open-source AI Operating System: your agents, their skills, your company memory, and every connector in one git repo you own, with the agents working on real cloud computers.
What OpenAI Dots are
Dots are OpenAI's always-on agents, and OpenAI describes them as agents that keep working on your behalf between conversations, which is the part that separates a dot from a chat window (introducing dots).
Each dot runs on its own cloud computer with its own browser, is powered by GPT-6 Astra, and reaches more than 4,000 apps through plugins. You can open that computer whenever you like to watch the work, and take over when you want to do a step yourself (introducing dots, help center).
You reach a dot in ChatGPT on desktop, web and mobile, and in Slack and Teams. Texting is a limited beta for Pro plans in the US (help center).
How to use a dot
Setup runs in the ChatGPT mobile app, the desktop app or desktop web: create the dot, give it a name, connect the apps it should use, and let it introduce itself. Messaging channels such as Slack are connected from a desktop session as well, and mobile web does not support dots (help center).
The controls decide what a dot may do without you. Custom Rules set one of four behaviors for an action: take action without asking, take action if pre-approved, ask before taking action, or hand off to you. OpenAI's launch page describes the same controls as allow, require approval or block, and built-in safety requirements always apply (introducing dots, help center).
Activity View and the profile's In progress, Scheduled and Completed lists show what the dot did while you were away. Pause stops a dot until you resume it, and Reset deletes the dot with its conversations, saved memories and scheduled tasks (help center).
Availability is gated by plan and market. Pro is open to people 18 and older outside the European Economic Area, Switzerland and the UK. Business Premium depends on the workspace's region and eligibility, and Enterprise, including Edu and Healthcare, is a beta an admin enables (help center).
What a dot is not
It is not a computer you host. The cloud computer belongs to OpenAI, and your own machine stays separate unless you connect it, so you inspect the work rather than run the machine (introducing dots).
It is not a model you choose. Both OpenAI pages name one model behind every dot, GPT-6 Astra, and the documented setup offers no model picker (help center).
It is not a rule set you keep. Custom Rules sit under Customize in the ChatGPT app, and the Plugins screen uses shared ChatGPT settings, so the policy lives in an account rather than a file. Neither OpenAI page describes exporting a dot or its rules to another environment (help center).
What rules for an agent look like when the file is yours
# v2 is deny-by-default: an omitted connectors/secrets/kortix_permissions/# skills grant resolves to "none", so a scoped specialist starts locked down.agents: kortix: file: agents/kortix.md kortix_permissions: allSix lines from the manifest a Kortix project commits, read from the public starter template.
The same job with the computer and the rules in your repo
Kortix keeps the same ingredients somewhere you own them. The six lines in that manifest are the governance half of an agent: which agent exists, which markdown file holds its behavior, and how far its reach goes. That file sits in the project's git repo next to the code, so granting a connector is a diff a person reviews and the record of who widened what stays readable.
The deny-by-default comment earns its place. An agent with no grant resolves to none, and the starter's own agent is granted broadly on purpose, which is the setting a team narrows first. Because the file and the code are yours, where it runs is your call: run it on Kortix Cloud, in your VPC, or on your own on-prem network. Self-host is free. Open source. Read it, fork it, audit it.
What people ask about the word dot
The two questions a reader has about the name, and what it points at.
What are ChatGPT dots?
ChatGPT dots are the same product seen from inside ChatGPT: a named, always-on agent in your ChatGPT account with its own cloud computer, Custom Rules and an Activity View. Dots reach Pro plans outside the EEA, Switzerland and the UK first.
Is there a dot in AI?
Yes, in OpenAI's sense: a dot is one always-on agent with its own cloud computer. The word is OpenAI's product name, not a general term for a model or a chat window.
Put the computer and the rules in your own repo
One project, one manifest, and a rule file you can read.