OpenAI just dropped "Dots," its new always-on AI agents designed to proactively help users and make autonomous decisions online, directly challenging Meta's "Muse" and signaling a major shift from chatbots to personal AI assistants.
At OpenAI's recent DevDay, the company unveiled Dots, personal AI agents that operate continuously, scouring the web to complete tasks and even make independent choices. After witnessing demos and hearing CEO Sam Altman discuss their capabilities, the author tested out their own agent, named "Toolie" and styled as a pink frog, comparing it to Meta's Muse, which they had been using for a month.
Toolie, accessed via ChatGPT, proved its worth by identifying actionable tasks within the author's chat history. It offered to draft a follow-up email for a data request, suggested dinner spots for an upcoming trip, and proposed refining pitch drafts. While the author declined the last offer, the agent's ability to manage time-consuming tasks was evident.
This development signifies a broader trend in Silicon Valley, where companies like Instinct, OpenClaw, and Google are all investing heavily in AI agents. Dots and Muse are pushing these agents into the mainstream, promising to automate everyday chores like booking appointments and managing schedules. While current Dots are only available to OpenAI's Pro subscribers ($100/month), Meta's Muse remains free, highlighting different market strategies.
OpenAI emphasizes the advanced GPT-6 Astra model powering Dots, acknowledging the significant computational resources required. Sam Altman hinted at future mass-market availability, suggesting that the premium nature of the current offering is a strategic starting point. Both OpenAI and Meta are adopting "cute" aesthetics for their agents, like Muse's Labubu style and Dot's puffball design, potentially to foster user affection and disarm concerns about AI.
The "always-on" nature of these agents is a key feature, allowing them to initiate tasks and communicate proactively, unlike passive chatbots. Altman shared that his own Dot handles his mornings, reducing mental clutter. While the potential for helpfulness is high, the balance between proactive assistance and intrusive notifications remains a critical area for development, as seen when Muse reminded the author about a utility bill they didn't need prompting for.
Despite potential concerns about data privacy, such as granting access to Gmail and Google Drive, the author notes a significant improvement in agent reliability compared to a year ago. Earlier agents struggled with basic computer interactions, whereas current iterations show greater potential for completing assigned tasks. However, the author maintains a cautious approach, emphasizing the need for AI agents to earn user trust before handling critical aspects of life.
While OpenAI remains tight-lipped about future hardware integrations, the foundation laid by Dots suggests potential applications in AI-powered alarm clocks or voice assistants. As the competition in the AI agent space intensifies, Dots and similar technologies are poised to become indispensable tools, moving from niche Silicon Valley interest to everyday consumer use.
