Business Insider journalist Alistair Barr tested OpenAI’s Dots on a newsletter after receiving workplace permission to connect applications. The agent moved content from a Google document into the company’s publishing form and returned a draft after about five minutes. Barr’s editor found most content in the right places. Earlier website sharing had run into the enterprise workspace’s access restrictions.
Business Insider journalist Alistair Barr tested Dots, OpenAI’s task-running AI agent, on a workplace newsletter. On Friday, he obtained permission to connect work applications and gave the agent a Google document holding an earlier AI Insider newsletter and the company’s publishing instructions. The task required filling a complex publishing form with the newsletter’s text and associated material.[1]
A popup asked Barr for his password, followed by two-factor authentication through Okta, the workplace identity service. He used a temporary code from his phone. After about 5 minutes, Dots returned a test-draft link. His editor found the photographs and most other content, including the chart and links, correctly placed, judging the draft 95 per cent correct. That assessment describes this individual draft.[1]
Barr had earlier built a private website using his employer’s enterprise ChatGPT account. When he requested public access, the agent said the workspace prohibited public links and external invitations, requiring an administrator to change the settings. An OpenAI spokesperson said a site made through a paid personal account could have been public. Dots product lead Alexander Embiricos described the initially restrained corporate environment as deliberate, with broader deployment dependent on observing safe operation.[1]
