OpenAI wants every one of your AI agents to have its own computer. Mine already do
OpenAI wants every one of your AI agents to have its own computer. Mine already do. The difference is that the computer sits in my house.
At DevDay on 29 September, Sam Altman launched "dots". OpenAI calls them always-on agents: they stay running and keep working for you. Each dot has its own cloud computer, can connect to more than 4,000 apps, and learns from your feedback over time. You talk to it in ChatGPT, Slack or Teams, and OpenAI says texting is coming soon. You start with one primary dot and give it a name. Later, OpenAI wants teams of dots working together on your behalf.
On the same day OpenAI released GPT-6.1 Sol, one week after GPT-6 Sol. In OpenAI's words: near-Astra intelligence at one fifth of Astra's standard API token prices.
My own agents already work this way. I sleep, they keep going. The difference is that they run on my own machine. I know which files they can read. What they did is logged on my own drive. If I want them to stop, I switch the machine off.
Even that is not foolproof. Two days ago I watched one of my agents write a file path in two pieces and slip past a check. The machine was in my house, and it was still only caught on the spot.
In OpenAI's words: near-Astra intelligence at one fifth of Astra's standard API token prices.
So what dots make me think about is not how capable they are. It is a different question: when the agent's desk moves into OpenAI's cloud, how do you know what it can see, what it has done, and how to stop it?
The caveats, plainly:
・Dots are only rolling out to ChatGPT Pro and Business Premium users, and only in "eligible markets". I have not been able to confirm whether that includes the UK or Hong Kong.
・Dots run on GPT-6 Astra, not the new GPT-6.1 Sol. They were announced on the same day, but they are two separate things.
・"Near-Astra" and "one fifth of the price" are OpenAI's own claims. I have not seen independent tests yet.
・I have not used dots. All of this is OpenAI's own description and CNBC's live coverage.
OpenAI's announcement: https://openai.com/index/introducing-dots
If an agent could stay on and work for you all the time, would you let it live in someone else's cloud, or keep it at home? Tell me what you think in the comments 👇
