On September 29, OpenAI introduced Dots at the DevDay event. Testers note that the agents are capable of sorting emails, identifying schedule conflicts, and reducing office workload. Each Dot operates on a separate cloud computer, allowing users to interact with it even when the laptop is closed. Through ChatGPT, the agent connects to more than 4,000 applications using plugins chosen by the user.
One tester, who used Dots for several days before launch, named his agent Boo. He granted it access to his email and Slack. Boo filtered incoming requests, allowing them to be sorted without opening the email, and acted without direct instructions: when the tester rescheduled a flight, the agent noted a conflict with an unread meeting request in Slack. Another tester, a tech journalist, assigned his Dot a range of office tasks. In his assessment, the agent declined a radio invitation, answered accountant queries, and prepared a meeting agenda from a transcript. The journalist estimated that Dot completed about two hours of work in 15 minutes of his time.
However, the first tester warned of errors in the pre-launch build: issues with permissions, lost messages, and lack of iMessage support. He advised most users to wait one to two weeks. Comparing Dots with Meta's Muse and SpaceXAI's Grok Bot, he noted that users already actively using these tools might not rush to switch. Another tester drew a sharper line: Dots seemed more work-oriented to him, with Slack being the strongest channel. As a personal assistant, he still ranked Grok Bot higher, partly due to the less intuitive plugin setup. Among the advantages, he noted the ability to call the agent – this allowed communication on the way to work. Payments, he said, still do not work properly, despite the on-stage example of ticket purchases by OpenAI CEO Sam Altman.
Related articles
