When you connect an assistant to your email, calendar, or online drive, you give it a way to act in those apps. The agent does that work through your account, with your name on it.
It Acts as You
Connecting an app usually means signing in and clicking Allow. From then on, the agent can do whatever that connection allows, like reading, sending, or editing, and it does it as you. An email it sends comes from your address. A calendar invite comes from your calendar. An order goes on the company card saved in your account.
Websites and apps usually can’t tell the agent apart from you. Some agents don’t tell websites they’re agents, so to the site they look like a person using a browser. When something goes wrong, the people on the other end see your name on it.
Asking Before It Acts
Before an agent does something that’s hard to take back, like sending, buying, or deleting, the app usually stops and asks you first. Those prompts come from the app, which checks with you before it carries out what the model asked for in the agent loop. They’re your chance to read what’s about to happen and say no.
