OpenAI turns its coding tools into «dots», an always-on personal assistant

OpenAI, the US artificial intelligence company, unveiled a personal assistant called «dots» at its developer day in late September. The tool keeps working in the background even when the user isn't asking it to do anything. Alexander Embiricos, who leads OpenAI's enterprise product team, described where the product came from at the TechCrunch Disrupt conference this month. Always-on assistants have become one of this year's main answers to how people will use deep learning models; Meta offers a comparable product called Muse.

How it works

The assistant is available in the ChatGPT desktop and mobile apps for subscribers on the Pro tier or higher, with plans starting at $100 a month. Once connected to a user's calendar, email, computer and other tools, it tries to help get work done. Embiricos noted that much of the same functionality is possible with Codex, OpenAI's coding tool released at the start of this year, but that requires more hand-holding. The assistant grew out of the agentic tools built for software engineers, which the company had already adapted into ChatGPT Work.

A feature called «proactive research» has the assistant periodically review what it knows about the user and what it can access, then investigate how it could be more helpful and propose actions. In TechCrunch's testing, it flagged calendar conflicts and deadlines, though not always in good time.

Privacy and scale

OpenAI says the assistant has the same privacy controls as ChatGPT: users can turn off model training on their conversations and control what the assistant can access on their computer. The controls are not granular, however, and the app refuses, for example, to accept read-only access to cloud drives. Embiricos acknowledged the need for much finer controls and said the conservative default settings at launch drew criticism.

He also said OpenAI is in many conversations with payment processors, but declined to give further detail.