OpenAI's New Dots Agents Feel Like Office Software That Can Also Order Lunch

OpenAI has introduced Dots, a new agent platform built around small, blobby, anthropomorphic avatars that users can name and customize. The company announced it earlier this week, and the early impression is that the cute design hides a product aimed squarely at the workplace.

The comparison that comes up first is Meta's Muse, which also uses a character-driven approach to AI assistance. Where Muse is built to feel approachable and casual, Dots feels closer to enterprise software. It can still do everyday personal tasks, such as ordering you dinner, but the overall experience reads as a work tool that happens to be capable of handling a burrito order.

At the moment each user gets a single Dot. OpenAI says that in the future people will be able to run several at once, which suggests the company sees the product as a team of specialized helpers rather than one general assistant. The reviewer who tested it named theirs Dotty McDotface, a reminder that the personality layer is meant to be playful even when the tasks are not.

The interface follows a familiar pattern for agent products. You chat with your Dot in one window, and in another you can watch it work through the task as it goes. That side-by-side view is meant to make the agent's actions visible, which matters for anything that touches real accounts, payments or company data.

The launch fits a broader race among AI companies to turn chatbots into agents that act on a user's behalf. Meta is pushing a consumer-friendly angle, while OpenAI appears to be prioritizing business buyers and using the avatar design as a way to make autonomous software feel less intimidating.

Whether the mascot approach helps or distracts in a professional setting remains to be seen. For now, Dots shows how OpenAI is positioning its agents: not as novelty companions, but as workplace tools with a friendly face and a few personal-assistant tricks on the side.