This feature allows teams to seamlessly combine AI automation with human judgment during live conversations. You can step in when needed, take full control, and then hand the conversation back to the AI—without losing any context.
The result: smoother conversations, better outcomes, and full control at critical moments.
What You Can Do Inside a Conversation
From now on, within each active conversation, you can:
Intervene in real time while the AI agent is interacting with the user.
Take full human control whenever the situation requires a personal touch.
Reactivate the AI agent and let it continue the conversation seamlessly, with full context preserved.
This hybrid approach ensures you get the efficiency of automation without sacrificing quality or empathy.
When Should You Intervene?
Human intervention is especially useful when:
A user explicitly asks to speak with a human.
The conversation becomes sensitive, complex, or strategic.
You want to close a deal, handle objections, or provide tailored advice.
The AI needs guidance to continue in the right direction.
Think of the AI as an excellent first responder—and humans as the specialists who step in at the right moment.
How the Handoff Works
AI is active and handling the conversation automatically.
You intervene and start responding manually in real time.
The conversation clearly reflects that a human has taken over.
When ready, you reactivate the AI, and it continues exactly where the conversation left off.}
Why This Matters
By blending AI and human control, you achieve:
More natural and fluid conversations
Higher user trust and satisfaction
Faster resolution times
Better business outcomes
Automation handles scale. Humans handle nuance. Together, they perform better 🤝
Proactive Human Intervention with Alerts
To further optimize human time and attention, you can define agent prompts and rules that automatically detect moments where human intervention is required.
For example, the AI agent can be instructed to:
Detect user friction, confusion, or negative sentiment.
Identify explicit requests such as "I want to talk to a human" or similar phrases.
Recognize high‑risk or high‑value scenarios (complaints, cancellations, pricing discussions, or sales opportunities).
When any of these conditions are met, the system can automatically notify a human via Slack, email, or another internal channel, prompting them to step in and take control of the conversation.
This means:
Humans don’t need to monitor every conversation in real time.
Intervention happens only when it truly matters.
Response times improve without increasing operational overhead.
The AI handles the flow. Alerts handle prioritization. Humans step in exactly when needed.
Best Practices
Intervene early when you detect friction or confusion.
Keep responses clear and aligned with the brand tone.
Hand control back to the AI once the critical moment has passed.
Use human intervention strategically—not reactively.
If you have any questions about this feature or best practices, feel free to reach out to the Help Center or your account manager.