> For the complete documentation index, see [llms.txt](https://docs.patagon.ai/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.patagon.ai/en/sales-performance/how-human-handoff-works.md). # How Human Handoff Works For rating definitions, see the [ILA lead rating system](/en/agent-optimization/ila-rating-system.md). #### What is human handoff Human handoff is the mechanism by which a team member temporarily takes control of a conversation being managed by the AI agent. During the handoff, the agent is **completely silenced** for that specific conversation: it doesn't send any messages until control is manually returned. {% hint style="info" %} **Path in the app:** Leads → Conversations → open a conversation → **Take Control** button (bottom of the panel). {% endhint %} #### What happens to the agent during handoff This is the most common question, and the answer is straightforward: **the agent completely stops responding in that conversation.** As soon as you click **Take Control**: * The input field is unlocked for you to type freely. * The AI agent **doesn't send any messages**: neither replies to the lead's questions nor automated follow-ups. * Two icons appear in the conversation header: a **robot (gray/disabled)** indicating that the agent is paused, and an **exit icon** to return control. * In the Conversations list, the lead **doesn't disappear**: it stays visible to all members with access. {% hint style="warning" %} Handoff is **per conversation, not global.** If you take control of Lead A's conversation, the agent keeps responding normally for every other lead. Only the conversation where you clicked "Take Control" is paused. {% endhint %} #### How the agent resumes the conversation after handoff The resume process is simple and immediate: 1. When you're done, click the **exit icon** in the conversation header (or the **Leave conversation** button). 2. Control returns to the AI agent instantly. 3. The agent resumes automated responses for that lead from that moment on. 4. The input field is locked again for the human, showing the default state once more. **What the agent "remembers" when resuming:** It has access to the full conversation history, including the messages sent by the human during the handoff. Qualification continues from where it left off, taking into account everything that was said. {% hint style="success" %} **Tip:** before leaving control, add an **Internal Note** (Notes tab, in Lead Details) explaining what was discussed and what the expected next step is.\ This guides the agent and the other team members. {% endhint %} #### The follow-up issue during handoff (recurring bug) This is one of the points that causes the most confusion on Slack. Here's the contrast between the documented behavior and the reported behavior: {% tabs %} {% tab title="Expected" %} According to the official documentation, **the agent doesn't send any messages while the human is in control** — this includes automated follow-ups. The agent is paused. {% endtab %} {% tab title="Reported (bug)" %} Users report that the agent keeps firing the **Initial Delay follow-up** (e.g., 120 minutes) even when a human is in control of the conversation. {% endtab %} {% endtabs %} **Why this can happen** The most likely cause lies in the **follow-up timing logic**. The system calculates follow-ups based on the **lead's last message**, not the handoff status. If a follow-up was scheduled before the handoff started, the system might fire it before checking if control was transferred. This timing gap causes the bug. ``` Lead sends a message at 10:00 ↓ Handoff started at 10:05 (human takes control) ↓ Follow-up timer: 120 min after the lead's message = 12:00 ↓ [BUG] If the system doesn't check the handoff state at the moment of firing, the follow-up goes out at 12:00 even with the human in control. ``` **How to mitigate while the bug isn't fixed** There are three practical approaches: {% tabs %} {% tab title="1 · Disable follow-up" %} Before taking control of a sensitive conversation, go to **Agent → Capabilities → Features → Follow-up** and turn off the ***Enable Follow-up*** toggle. Remember to re-enable it afterward. {% hint style="warning" %} Affects all leads, not just one. Use with care during peak hours. {% endhint %} {% endtab %} {% tab title="2 · Leave handoff quickly" %} The less time the handoff stays active, the lower the chance of the follow-up colliding. If the conversation is finished, leave immediately instead of keeping the handoff active for hours. {% endtab %} {% tab title="3 · Send a message" %} Sending any message during the handoff resets the 24h timer — which can "reset" the follow-up counter when the agent resumes control, avoiding the immediate firing. {% endtab %} {% endtabs %} {% hint style="danger" %} **Hit by this bug?**\ Report it to [*support@patagon.ai*](mailto:support@patagon.ai) with the lead ID and the time of the wrongful firing. Include screenshots of the conversation showing that the handoff was active and the follow-up was sent anyway. {% endhint %} #### How the 24h window behaves during handoff The 24h timer **keeps running** during the handoff — it doesn't pause. This has practical implications:
Window stateWhat happens
Active (time remaining)The human can type freely and send any message.
Expired (timer = 0)The conversation becomes inactive. The bottom field switches to a Template selector instead of free text, and only Meta-approved templates can reopen the conversation.
When the window expires, the robot icon's tooltip shows: > "This conversation is inactive. You won't be able to send more messages to this user unless they send a new message first." {% hint style="danger" %} **Critical warning:** if you take control of a conversation with less than 1 hour left on the timer, prioritize sending the human message immediately. If the window closes while you're "thinking about what to write," you'll have to use an approved template — which completely changes the approach. {% endhint %} #### Best practices: handoff vs. letting the agent continue **Take control when** **By lead urgency or quality** * The lead is **5★ (Qualified + Meeting)**: respond within 5 minutes. Human response speed at this moment directly impacts conversion. * The lead is **4★ or 5★ with negative sentiment**: needs objection handling that goes beyond the agent's script. * The lead has **confirmed budget and decision authority** and is ready to close: the ideal moment for a human to lead. **By window limitation** * The timer is **below 1 hour** and the conversation is still open: act before the window closes. * The lead replied to a post-24h follow-up template and showed real interest: take advantage while the window is reopened. **By the nature of the conversation** * The lead asked a very specific question about product, price, or terms that the agent isn't trained to answer correctly. * There are signs of frustration or complaint: immediate human intervention prevents churn. * The lead explicitly asked to speak with a person. **Let the agent continue when** * The conversation is still in the **qualification phase** (answering initial questions, collecting data). * The lead is **1★ or 2★**: a 1★ lead has provided no useful qualification or ICP information; if they never replied, they may receive at most one light first follow-up. A 2★ lead is in an active qualification conversation but is not potential yet. Neither is ready for Sales handoff. * The agent is responding correctly and the lead is engaged: interrupting can create a discontinuity in the experience. * You won't be able to respond immediately: it's better for the agent to continue than for the lead to wait for a human who won't show up. *** #### Quick reference table | Question | Answer | | ------------------------------------------------- | ----------------------------------------------------------- | | Does the agent stop responding during handoff? | Yes — completely, for that conversation. | | Does the follow-up stop during handoff? | It should, but there's a documented bug of wrongful firing. | | Does the 24h timer pause during handoff? | No — it keeps running. | | How does the agent resume? | Click "Leave conversation". Resuming is immediate. | | Does the agent remember what was said in handoff? | Yes — it has access to the full history. | | Does handoff affect other leads? | No — it's per conversation, not global. | | How do I tell the handoff is active? | Gray robot icon + exit icon in the conversation header. | | Where is the "Take Control" button? | Leads → Conversations → bottom of the chat panel. | --- # Agent Instructions This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com. ## Querying This Documentation If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question. Perform an HTTP GET request on the current page URL with the `ask` query parameter, and the optional `goal` query parameter: ``` GET https://docs.patagon.ai/en/sales-performance/how-human-handoff-works.md?ask=&goal= ``` `ask` is the immediate question: it should be specific, self-contained, and written in natural language. `goal` is optional and describes the broader end goal you are ultimately trying to accomplish on behalf of the user. GitBook uses it to tailor the answer towards what is most useful for that goal. The response will contain a direct answer to the question and relevant excerpts and sources from the documentation. Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.