OPENAI LAUNCHES DOTS AGENTIC ASSISTANT AT DEV DAY

OpenAI announced a new agentic assistant called Dots at its Dev Day event, according to company statements. Dots are described as always-on AI agents powered by OpenAI's GPT-6 Astra model, designed to operate independently of any specific device and carry out user-defined tasks continuously in the background with minimal supervision. OpenAI says the assistants can "do nearly anything" by using a dedicated cloud computer to access a web browser and more than 4,000 connected apps. Users can interact with a Dot through a text-based interface or by voice call within ChatGPT on web, desktop or mobile. The assistant can also connect to Microsoft Teams and Slack, carrying context between sessions across different apps and devices.

OpenAI says Dots differ from the company's existing Codex and ChatGPT products by pursuing goals over extended periods rather than responding to single requests. Once assigned a task, a Dot sends updates and questions to the user while working, and its progress can be tracked through its cloud computer interface. OpenAI gave examples including a developer asking a Dot to build and test app updates based on customer feedback, and a content creator using one to review an interview transcript, create clips, write show notes and draft social media posts. The company says Dots learn from users' preferences over time, allowing them to personalise output and suggest further tasks. Users can also collaborate with a Dot alongside colleagues in a new area of the app called ChatGPT Space.

Coverage of the launch noted that Dots arrives shortly after Meta introduced a similar product called Muse, positioning OpenAI's assistant as a competitor in the emerging market for always-on AI agents. OpenAI currently limits users to creating one Dot each, though the company says it plans to allow multiple agents to be deployed at once in future. OpenAI also says Dots will soon be able to communicate with users over text message, in addition to the existing chat and voice options. No date has been given for when these additional features will become available.