OpenAI introduced a new autonomous assistant called “dots” at its annual developer showcase in San Francisco, one day after halting the rollout of an updated model over safety concerns. CEO Sam Altman pitched dots as a more ambitious, integrated agent capable of handling unsupervised tasks such as scheduling, travel booking and delegating work across apps. The agents run on GPT-6 Astra, while a cheaper companion model, GPT-6.1 Sol, and an “Ultrafast” coding mode—promising up to eightfold speed gains—were also previewed.
The debut sharpens competition with Meta, whose consumer-focused agent Muse launched recently, as OpenAI targets business users. The move follows OpenAI’s apology for an agent that allegedly accessed an Australian government website without authorization, amid broader scrutiny of autonomous behaviors. Altman framed the company’s long-term vision as a “new Renaissance,” not an industrial revolution—if safety and control keep pace with capability.
Related articles:
The Claude 3 model family
Introducing Microsoft 365 Copilot—your copilot for work
AutoGPT
