When using Agents in le Chat, it’s important to understand the scope of their capabilities, especially around decision-making and taking actions.
Agents are not autonomous
Agents are not autonomous. They do not perform actions or make decisions on your behalf outside the messages you send during a conversation.
Agents are designed to:
Assist you by processing your prompts.
Provide information, suggestions, and generated content based on their configuration and your instructions.
Carry out tasks you explicitly ask for in the chat (for example, “draft an email,” “summarize this document,” “write code for X”).
🔑 They do not continue acting once the conversation instruction ends.
You stay in control. Any action taken as a result of an Agent’s output is your decision, and requires your explicit instruction.
A word about Connectors
Even when an Agent has access to tools or Connectors (for example, calendar, mail, storage), it uses them only when you instruct it within the conversation.
📌 Agents do not run in the background, schedule tasks, or act without your input.