Why OpenAI’s Dots and Meta’s Muse Give AI Agents a Cute Face

OpenAI and Meta are giving personal AI agents a more familiar face: colorful, customizable characters users can name. The design can make software that acts on a person’s behalf feel less intimidating. But an appealing mascot is not evidence that an agent is safe, reliable, or suitable for access to sensitive accounts. 135

Dots and Muse: the agents behind the mascots

OpenAI’s Dots are described as always-on agents that can pursue tasks across apps. Reported examples include rescheduling appointments, booking flights, and updating code. Their avatars come in different colors and styles, and users can name them. 1519

Meta’s Jolly is the mascot for its Muse assistant, not a separate agent. Reporting describes Jolly as a customizable character with rosy cheeks. Muse was introduced before Jolly was added to its presentation. 3

The distinction matters: the character is the interface and branding; the agent is the software that may take actions. A charming avatar does not make those actions less consequential.

How the companies present their agents

Meta’s Muse promotion has used everyday-life scenarios. One reported television ad showed Muse helping with tasks such as sorting email, organizing a family calendar, and buying school supplies. Jolly gives the product a recognizable character, but was not featured in that launch ad, according to reporting.

OpenAI introduced Dots at its September 29, 2026 DevDay, presenting it as a competitor to Muse. Its launch materials showed customizable characters, while coverage emphasized that users could name their agents. 58

These examples show how both products are framed: not simply as chatbots, but as assistants that can take action. They don’t establish how much either company has spent on promotion or whether cute branding changes users’ behavior.

What Dots costs—and what the demo showed

At launch, OpenAI said a user’s first Dot was included with eligible Pro or Business Premium plans. Wired reported that the Pro plan started at $100 a month; the price of a plan should not be confused with a separate fee for the first Dot. Availability was limited to eligible markets. 11

Muse, by comparison, was free to use through its app or website, according to Wired. 11

Dots also had a rough moment during its live presentation: one agent failed while trying to carry out an assigned command, and another demonstration encountered a voice-command glitch. A stage failure does not by itself establish how the product performs in routine use, but it is a reminder that a polished product concept and dependable execution are different things. 20

Safety, trust, and what a mascot can’t prove

The timing added to questions about readiness. OpenAI launched Dots a day after reporting that it had held back a different system over safety concerns. The company said it had conducted additional safety testing on Dots; that statement is not proof that every action an agent might take will be safe. 717

A friendly character may affect what users expect from a product. Reporting on the launches raised the possibility that approachable design could make an agent seem safer or more dependable before people have much experience evaluating its work. That is a concern about perception—not evidence that Dots or Jolly have already caused users to share more data or become more attached to a company. 3

For users, the practical test is the agent’s behavior and controls, not its appearance. Before connecting accounts or assigning tasks, check what information the agent can access, what actions it can take, and when it asks for permission. Cute branding may make an agent easier to approach; trust still has to be earned through clear limits and reliable performance.