The word plugin is used loosely, but in AI tools it usually means an add-on that gives an assistant new abilities. Examples include connectors to a calendar or document store, a tool that runs a calculation, or a browser extension that reads the current page.

Three common types

  • Connectors: link the assistant to a service, such as a file store or project tracker, so it can read or change data there
  • Tools: give it a specific action, such as converting a file or querying a database
  • Interface add-ons: add buttons, panels or shortcuts to the product itself

Why they matter

An assistant is limited to what it can see and do. A plugin widens that boundary, which makes the assistant more useful, and also means it can reach more of your data.

Key idea

Every plugin is code or configuration that someone wrote, and it runs with the permissions you grant. Before installing one, know who wrote it, what it can access and how you will remove it.

Plugin support and naming differ between products, so read the documentation for the tool you use to see what is available.