OpenAI launched new autonomous agents called Dots

OpenAI introduced Dots, a new class of always-on autonomous agents, at its annual developer event. These software assistants are designed to proactively manage tasks, make independent decisions, and browse the internet to help users.

How the agents function

Unlike traditional chatbots that wait for user prompts, these agents act on their own. They monitor user history to suggest improvements, such as drafting email replies or optimizing calendar plans. The system uses the GPT-6 Astra model to power its reasoning capabilities.

Users customize their agents by setting preferences and providing access to digital accounts like Gmail and Google Drive. The interaction feels like a group chat, with the agent using emojis to confirm when it has completed a request.

Market competition

Dots enters a growing field of personal agents, including Meta’s Muse. While Muse is free, Dots is currently limited to subscribers of the OpenAI Pro plan, which costs $100 per month. Sam Altman indicated that a broader, mass-market version is part of the long-term plan.

Designers are using friendly, cute visuals to help users feel more comfortable with these persistent tools. The goal is to remove the burden of repetitive, manual tasks from the daily routine.

Current limitations

These agents are still being tested and do not always perform perfectly. Security remains a significant concern, as connecting personal data to an autonomous system carries inherent risks. The software also struggles with some complex interactions, and users should carefully decide which account permissions to grant.