What are ChatGPT dots?

ChatGPT dots are OpenAI’s latest step into always-on AI agents, following moves by Meta, Google, and Anthropic. These agents operate on their own cloud computers, allowing them to continue working even when a user’s laptop is closed. They connect to apps like email, calendars, and Slack, and can message users when they need input or detect urgent issues. The concept builds on prior coding agents but applies their capabilities to broader personal and work tasks.

Each dot runs on OpenAI’s GPT-6 Astra model and includes a Linux-based cloud computer with a Chrome browser, enabling web browsing, file creation, and tool use. Users can personalize their dot’s appearance and name, and can observe or take over its actions, such as entering passwords or adding payment methods. The agent pings users before major actions like sending messages or using payment cards, though frequency depends on user settings.

Dots leverage ChatGPT’s existing plugin system, so prior app integrations carry over automatically. Users can connect services like Gmail, Google Calendar, Slack, and over 9,000 others via Zapier without reconfiguring permissions. Dots are multimodal, handling photos, screenshots, and documents, and can perform proactive research by scanning connected apps for ways to assist without explicit instructions.

In testing, dots flagged bounced payments, spotted delivery conflicts, and updated calendars automatically. However, they occasionally missed urgent items requiring high judgment while focusing on routine tasks. Deep integration with personal accounts also raises privacy and security concerns, as dots require broad access to monitor and act across multiple services.