Particle.news

OpenAI Launches Dots, Always-On Personal Agents

The agents run on GPT-6 Astra in isolated cloud 'computers', carry context across apps, and are rolling out to paid ChatGPT tiers with built-in read-only research and an auto-review safety workflow.

Overview

  • OpenAI unveiled Dots at its DevDay on Tuesday, September 29, 2026, calling them persistent assistants that continue work without new prompts and can be messaged from ChatGPT, Slack, and Microsoft Teams.
  • Dots run on GPT‑6 Astra in their own cloud 'computers' and browsers and can connect to more than 4,000 apps through OpenAI’s plugin ecosystem to carry context between tools.
  • OpenAI says background tasks are limited to read-only 'proactive research' while any action that could change accounts or share data goes through an 'auto-review' step governed by Custom Rules and monitoring.
  • The product is rolling out now to ChatGPT Pro and Business Premium subscribers in eligible markets with one Dot included, and admins can enable beta access for enterprise workspaces as specialist Dot pilots and Microsoft Agent 365 integration begin.
  • The launch follows a wider move toward always-on agents this year and could shift how teams automate multi-step work, but OpenAI warns Dots can err and recommends human review for consequential tasks.