AI Agents · Lesson

Human Handoff Protocols

Detecting handoff triggers and smoothly transferring to live agents.

When Should an Agent Hand Off?

Not every conversation should be handled by an AI agent end-to-end. Knowing when to hand off is as important as knowing how to answer. Common triggers:

Customer explicitly asks for a human
Detected anger or distress
Complex or ambiguous situation outside the agent's scope
Legal, safety, or compliance sensitivity

Detecting Handoff Triggers

Use an LLM classifier to detect handoff triggers in real time. Check on every agent turn — not just the first message.

import openai, json

client = openai.OpenAI(api_key='YOUR_OPENAI_KEY')

def should_handoff(message: str, history: list[dict]) -> dict:
    context = '\n'.join(f"{m['role']}: {m['content']}" for m in history[-4:])
    prompt = (
        f'Conversation context:\n{context}\n'
        f'Latest message: "{message}"\n'
        f'Should this be handed to a human agent? Reasons: '
        f'angry_customer, explicit_human_request, complex_issue, legal_risk, other.\n'
        f'JSON: {{"handoff": true/false, "reason": "..."}}'
    )
    resp = client.chat.completions.create(
        model='gpt-4o-mini',
        messages=[{'role': 'user', 'content': prompt}],
        response_format={'type': 'json_object'}
    )
    return json.loads(resp.choices[0].message.content)

All lessons in this course

← Back to AI Agents