Dots: From Answering Questions to Taking on Tasks
A first encounter with Dots: a personal agent with a computer, a browser and ongoing work to do.
What's the Point?
Originally published in Skip Intro on 2 October 2026. Time references and product descriptions reflect that edition.
Sam Altman takes the stage and unveils the next big thing: Dots, a personal agent that lives inside the chat.
This isn't just another chatbot that waits for a question and then answers it. Think of it more as a "digital worker" you can hand ongoing responsibility to. Every Dot has its own computer and browser in the cloud, so it can keep researching, gathering information, drafting a document, following up on a task or working with connected apps, even when we're away from the screen and our own computer is switched off.
The novelty isn't only that it can write or summarise. It can take on a big goal, say "track what customers are saying about the product, find the problems that keep coming up and send me a weekly update", and carry out a sequence of steps over time, without needing a reminder each time. It's still supposed to stop and ask for permission before sensitive actions, but the idea is to move from AI that answers questions to AI that takes on tasks and actually moves them forward.
The moment I switched it on, it rummaged through my inbox, quickly found a few urgent tasks I'd forgotten about, and set off to deal with them.
