OpenAI launches always-on Dots agents to rival Meta's Muse

OpenAI introduced Dots at its DevDay keynote on Tuesday, presenting them as always-on AI assistants and a response to Meta's buzzy Muse AI. The agents are represented by cute, customizable, bubbly, blobby avatars.

The Dots are built to handle work tasks on their own, such as fixing a bug reported in Slack or sending out a forgotten invoice, sometimes before anyone asks. Alongside the agents, OpenAI is releasing GPT-6.1 Sol, a cheaper model, while holding back its top-end GPT-6.1 Astra over safety problems.

The launch also sparked online chatter suggesting that Elon Musk's xAI had trolled the Dots rollout. Musk is a former OpenAI founder who left the company and launched competitor Grok.

These elements show how agent systems are being presented as a step beyond chat interfaces: instead of waiting for a user prompt, an always-on assistant is framed as a persistent digital worker that can observe workplace signals and take action. The examples given—bug fixes and invoices—point to the kind of routine operational bottlenecks that developers, product teams, and small businesses often face when tools, notifications, and deadlines are spread across different platforms. If the vision works as described, the appeal is not just generating text, but reducing follow-up effort and closing small loops that interrupt productivity.

The avatar design also matters. A customizable, bubbly, blobby visual identity can make an autonomous system feel more approachable, especially when the agent is allowed to act without an explicit command. Friendly characters may lower hesitation around automation, but they can also raise questions about how much control, visibility, and human review users expect when an assistant starts handling tasks on its own. The balance between convenience and oversight is likely to become a central part of public discussion.

The mention of GPT-6.1 Sol and GPT-6.1 Astra adds another layer: the launch is not only about agent behavior, but also about model strategy. Releasing a cheaper model can suggest an effort to make broader use more practical, while withholding a top-end model over safety problems signals that capability and caution are still being weighed. In agent contexts, safety concerns are often amplified because the system can move from suggestion to action, whether that involves software changes, communications, or financial follow-ups.

The competitive framing further shapes the story. Presenting Dots as a response to Meta's Muse AI places the release inside a broader race to define how everyday work may be assisted by AI. The reference to xAI and Grok shows that the rollout also carries social and reputational dimensions, since OpenAI's ecosystem and its rivals are often discussed through public commentary. What remains notable is the emphasis on proactivity: the assistant is described not merely as answering questions, but as noticing unfinished work and stepping in before someone asks.

In short, the launch combines three themes: persistent AI assistants, tiered model releases, and a public competitive landscape. The concrete examples give the concept a practical shape, while the safety note and avatar design hint at the tensions that come with giving AI agents a more active role in daily operations. The story is therefore less about a single novelty feature and more about how autonomous assistants may gradually be normalized inside workplaces, chat platforms, and the broader AI market.