A key feature of the LeadClose AI Agent system is its ability to recognize when it can't help and seamlessly hand the conversation over to a human agent. This ensures that the customer never gets stuck in a frustrating "chatbot loop."
When Does an AI Agent Hand Off a Conversation?
An AI agent will automatically un-assign itself and return a conversation to the main "Unassigned" queue under several conditions:
- No Relevant Information: If the user asks a question and the agent's RAG system cannot find any relevant information in its knowledge base to formulate an answer, it will hand off the chat.
- User Asks for a Human: If the user's message contains phrases that explicitly ask to speak to a person (e.g., "talk to a human," "can I speak to an agent?", "connect me to support"), the AI will immediately hand off the conversation.
- User Expresses Frustration: The AI is trained to detect sentiment. If a user expresses significant frustration (e.g., "This isn't helping," "You don't understand," "Useless bot"), it will proactively hand off the conversation to de-escalate the situation.
The Handoff Process
When a handoff is triggered, the following happens:
- The AI agent un-assigns itself. The conversation's assignee is set back to "Unassigned."
- The conversation appears in the main inbox queue, visible to all human agents.
- The AI leaves an internal note. This is the most crucial step. The AI agent will automatically add a private note to the chat timeline, visible only to your team, explaining why it handed off the conversation.
This internal note gives the human agent who picks up the chat immediate context, so they don't have to ask the customer to repeat themselves. They can see what the user asked and why the AI couldn't help, allowing them to provide a seamless and efficient support experience.