ChatGPT’s official account says dots, OpenAI’s always-on personal agents, can now follow custom instructions. The post also says they now give faster replies, use less jargon, and have better follow-through. You can tell your dot in the thread how to behave, or which actions need your go-ahead, and you can save those instructions in settings.
Your dot now has… Faster replies. Less jargon. Better follow-through. And, they can now follow your custom instructions. Tell your dot in the thread things like how to behave or which actions need your go-ahead. “always talk in lowercase.” “Show me a draft before sending any emails.” You can also set custom instructions and rules in Customize > Behavior on mobile, or through your dot’s menu on web and desktop.
November 12, 2026
The examples in the post are “always talk in lowercase” and “Show me a draft before sending any emails.” The official 16-second video walks through the desktop menu: under the ••• control on the dot’s profile, alongside Rename, Pause, Reboot and Delete, there is a Custom instructions item. That panel has a 35,000-character limit. While it’s empty, it suggests “ask before sending emails” and “never make purchases above $100”. The clip then types “always use lower case” and “for really urgent updates, make sure you notify me in slack.”
A note at the bottom of that panel reads: “These instructions build on the dot defaults. Your dot may update its instruction based on your chats.” The ChatGPT post says you can also set custom instructions and rules in Customize > Behavior on mobile, or through the dot’s menu on web and desktop.
Instructions, and the custom rules already in the docs
OpenAI’s Control your dot guide already says to start with clear instructions in the conversation. That’s also where it says to put preferences such as writing style, or how you like updates. Custom rules are a separate, optional control. They set when a dot can take a given action. On the desktop app and on the web, those rules sit under Settings > Personalization. On iOS and Android, they’re under Customize > Custom rules. Each rule names an action, then you pick one of four behaviours: take action without asking, take action when you say so, ask before taking action, or hand the step off to you.
The same docs, and the getting-started help page, spell out the limits. Instructions and custom rules “don’t grant access to an app or computer, override built-in safety requirements, or remove required confirmations such as approval to use a saved login.” Asking a dot to draft replies does not give it permission to send them. Some steps, such as changing a password, stay with you. “Your dot can make mistakes, including when following your rules.”
| Control | Where it lives | What it’s for |
|---|---|---|
| Custom instructions | In the thread, or, according to the ChatGPT post, Customize > Behavior on mobile and the dot’s menu on web and desktop. The official video shows a 35,000-character Custom instructions panel. | Standing guidance such as tone, or “show me a draft before sending emails.” The panel says these build on the dot’s defaults, and that the dot may update its instruction from your chats. |
| Custom rules | Settings > Personalization on desktop and web; Customize > Custom rules on iOS and Android, per OpenAI’s docs. | A named action plus one of four choices: act without asking, act when you say so, ask first, or hand it off to you. |
Who can use a dot
OpenAI’s current getting-started help page says dots are rolling out to Pro users aged 18 or older in markets excluding the European Economic Area, Switzerland and the UK, and to Business Premium users in supported ChatGPT regions, subject to workspace eligibility. For eligible Enterprise, Edu and Healthcare workspaces, an admin preview runs from 4 November through 18 November 2026; from 18 November, dots become available by default unless an admin opts out. Workspaces with UAE data or inference residency don’t get dots. Each member can create one personal dot. For personal and Business Premium accounts, access may take several days to arrive.
OpenAI has also recently added ways for a dot to run commands on a computer you connect, and to keep more than one local computer attached to the same agent.
Discover more from Sulat AI & Tech News
Subscribe to get the latest posts sent to your email.
