How to Set Up Chatbot to Human Agent Handoff
Why Handoff Matters
No chatbot answers every question perfectly. Customers get frustrated when an AI keeps looping on a question it cannot answer, giving generic responses, or asking them to rephrase when the real problem is that the question requires human judgment. A well-configured handoff catches these moments and moves the customer to a human agent before frustration sets in.
The goal is not to eliminate AI from the conversation. The chatbot handles the first 70-80% of questions that have clear, documented answers. Handoff exists for the remaining 20-30% where the customer needs a person, whether because the issue is complex, emotional, or outside the knowledge base.
How to Configure Handoff
In the AI Chatbot app, link your chatbot to a live operator account. This is the agent (or team of agents) who will receive conversations when the bot hands off. Each chatbot can connect to one operator queue, and multiple agents can share that queue through the multi-agent inbox.
Configure when the chatbot should offer handoff. Common triggers include: the customer explicitly asks for a human ("let me talk to a person"), the AI confidence score drops below a threshold, or the conversation has gone back and forth more than 3-4 times without resolution. You can also set specific topics that always trigger handoff, like billing disputes or account cancellations.
Create a clear message the chatbot sends when handing off: "I'm connecting you with a support agent who can help with this. They'll have our full conversation, so you won't need to repeat anything." Avoid vague messages like "please hold" with no context. Tell the customer what is happening and set expectations for wait time if possible.
The human agent should see the entire chatbot conversation when they pick up the handoff. This includes the customer's original question, any information the bot collected (order number, account details, issue description), and the reason for handoff. The agent reads the history and picks up where the bot left off. See How to Keep Conversation History Across Channels.
When no human agents are available, the handoff flow needs a fallback. Options include: collecting the customer's contact information and creating a support ticket, offering to continue via email, or letting the chatbot attempt a broader answer with a disclaimer. Never leave the customer in a dead-end where the bot says "transferring you" and nobody picks up.
Common Handoff Triggers
Explicit Requests
The simplest trigger: the customer says "talk to a human," "speak to an agent," "real person," or similar phrases. Configure your chatbot's system prompt to recognize these requests and immediately initiate handoff instead of trying to answer the question itself. This should never be blocked or delayed.
Repeated Failures
If the chatbot gives two or three responses that do not resolve the issue (the customer keeps asking the same question in different ways, or says "that's not what I asked"), trigger handoff automatically. The AI can detect this pattern through conversation length and sentiment analysis. Continuing to loop makes the customer more frustrated with each response.
Sensitive Topics
Some topics should always go to a human regardless of whether the AI could technically answer them: billing disputes, refund requests above a certain amount, complaints, legal questions, account security issues. These require human judgment and empathy that an AI cannot reliably provide. Configure these as keyword-based triggers in the chatbot settings.
Customer Sentiment
When the customer's tone shifts to frustration or anger, handoff early. Phrases like "this is ridiculous," "I've been trying for an hour," or strong negative language are signals that the AI interaction is failing and a human needs to step in. See How to Handle Angry Customers With AI Support.
Set up seamless chatbot to human handoff so customers always reach the right support, whether AI or a live agent.
Get Started Free