Core Concepts
Always-On Agent in plain English.
Also known as: persistent agent,ambient agent,standing agent,Dot
The one-sentence version
An AI agent that holds an ongoing responsibility and keeps working between your messages, rather than answering one prompt and stopping.
An always-on agent is given a job, not a question. It runs continuously on its own compute, watches for events such as new emails, tickets, or calendar changes, acts within the permissions it has been granted, and reports back, so it keeps working after you close your laptop. OpenAI's Dots, launched at DevDay 2026, are the clearest mainstream example: each runs on dedicated cloud compute, connects to thousands of apps, and is reached from ChatGPT, Slack, or Teams. Meta Muse, xAI's Grok Bot, and Claude's merged chat-and-Cowork experience are variations on the same idea. The shift from chat to standing responsibility changes the questions you ask of a product: what can it access, how does it ask before acting, where does it log what it did, and what does it cost to leave running. It also raises the stakes of permission scoping and isolation, which is why vendors pair these agents with confirmation steps and audit trails.