OpenAI introduced its latest AI agent, Dots, at the annual DevDay conference in San Francisco on Tuesday, but the live demonstration revealed significant issues. Holly Li, a member of the OpenAI product team, tried to showcase Dots’ capabilities during the event. However, after initiating a call with the AI, there was an agonizing ten seconds of silence before it failed to respond, leaving the audience in an uncomfortable pause.
“Can you catch me up on just user testing from last night?” Li asked while attempting to interact with Dots, affectionately nicknamed “Dottie.” After the lack of response, she quipped, “I guess Dottie’s having a slow morning,” in an effort to ease the tension as the audience watched. This moment highlighted the challenges OpenAI faces in marketing Dots as an “always-on” assistant designed to handle tasks efficiently in the background.
Powered by the GPT-6 Astra model, Dots is intended to assist users across various professional settings. According to OpenAI’s blog post, its capabilities include monitoring user feedback, creating bugfixes, learning research, running new analyses, and even automatically inserting new figures into academic papers. These features indicate a broad range of potential applications for professionals, from app developers to scientists.
Dots represents OpenAI’s answer to Meta’s Muse AI agent, which is freely available and aimed at everyday users on platforms like Instagram and Facebook. Dots is currently available only through a $100 per month subscription. OpenAI may debut new features in its paid tiers before making them more widely available, suggesting this could be part of a longer-term strategy.
The demonstration failure has sparked discussions about the reliability of advanced AI systems in real-time applications. This incident emphasizes the necessity of thorough testing and fine-tuning before products are presented on stage, especially when marketed as tools to enhance everyday productivity.
