> For the complete documentation index, see [llms.txt](https://docs.maiagent.ai/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.maiagent.ai/maiagent-user-guide/en/serve/copilot.md).

# Customer Service Copilot

Support Copilot: enable AI Copilot on a conversation platform so Agents draft reply suggestions for customer messages, and human support agents confirm or edit them before sending.

Support Copilot lets an Agent draft replies behind the scenes for human support agents: when a customer sends a message, the AI generates an "AI suggestion" that only support agents can see. Agents can send it as is, edit it before sending, or ask the AI to regenerate it. It suits support teams that want to adopt AI but still want a human to review every reply.

AI suggestions are generated by the Agent bound to the conversation platform. They use that Agent's role instructions, knowledge bases, and tools, so the quality matches replies the AI sends directly.

This page has two parts: [Enable AI Copilot](#enable-copilot) is set up by the builder in the conversation platform settings in MaiAgent Build; [How Support Agents Use It](#use-suggestions) explains how support agents generate, edit, and send AI suggestions in a conversation.

{% hint style="info" %}
AI suggestions are never sent to the customer. Content reaches the customer, under the support agent's identity, only after the support agent clicks Send.
{% endhint %}

## Who Can Use It <a href="#permissions" id="permissions"></a>

| Task                                               | Required permission                                                                                                                                  |
| -------------------------------------------------- | ---------------------------------------------------------------------------------------------------------------------------------------------------- |
| Turn on or adjust AI Copilot                       | <mark style="color:blue;">Conversation platform permission</mark>, with edit permission on that conversation platform                                |
| Generate and send AI suggestions in a conversation | <mark style="color:blue;">All conversations permission</mark>, and the user's role must be assigned to the Agent bound to that conversation platform |

For how to set up roles and permissions, see [Role and Permission Management](https://docs.maiagent.ai/maiagent-user-guide/en/org/roles/role-permission.md).

## Enable AI Copilot <a href="#enable-copilot" id="enable-copilot"></a>

<mark style="color:blue;">AI Copilot</mark> is a tab on the conversation platform settings page, configured separately for each conversation platform; for the structure of the settings page, see [Conversation Platform Settings Overview](https://docs.maiagent.ai/maiagent-user-guide/en/build/conversation-settings/platform-settings.md#tab-map). Website, LINE, Messenger, Instagram, Telegram, Microsoft Teams, WhatsApp, Email, Slack, LINE WORKS, Team+, Viber, and other platforms all have this tab.

{% stepper %}
{% step %}

### Open the conversation platform settings <a href="#open-inbox-settings" id="open-inbox-settings"></a>

From the left menu, go to <mark style="color:blue;">Customer Conversations</mark> → <mark style="color:blue;">Conversation Platforms</mark>, find the platform you want to configure in the list, and click <mark style="color:blue;">Edit</mark>. For other ways to get there, see [Open the Settings Page](https://docs.maiagent.ai/maiagent-user-guide/en/build/conversation-settings/platform-settings.md#open-settings).
{% endstep %}

{% step %}

### Switch to the AI Copilot tab <a href="#open-copilot-tab" id="open-copilot-tab"></a>

Select the <mark style="color:blue;">AI Copilot</mark> tab (for the website platform, it is in the <mark style="color:blue;">Advanced Features</mark> group in the left navigation). This tab appears only after the platform has been created.
{% endstep %}

{% step %}

### Turn it on and choose a trigger mode <a href="#choose-trigger-mode" id="choose-trigger-mode"></a>

Turn on <mark style="color:blue;">Enable AI Copilot</mark>, choose when suggestions are generated under <mark style="color:blue;">Trigger Mode</mark>, and finally click <mark style="color:blue;">Save</mark>.

<figure><img src="https://1360999650-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2F6v6TNkkOQVfRYfcNirHL%2Fuploads%2Fgit-blob-a960a8ebd0a93441c17f2248a1693cf0681377e1%2Fcopilot-settings-tab.png?alt=media" alt="The AI Copilot tab on the conversation platform settings page: the Enable AI Copilot switch is on and the trigger mode is set to Manual"><figcaption><p>AI Copilot tab: turn it on, choose a trigger mode, then click Save</p></figcaption></figure>
{% endstep %}
{% endstepper %}

### Two Trigger Modes <a href="#trigger-modes" id="trigger-modes"></a>

| Mode                                                    | When suggestions are generated                                                                 | Does the Agent reply to the customer directly?                            |
| ------------------------------------------------------- | ---------------------------------------------------------------------------------------------- | ------------------------------------------------------------------------- |
| <mark style="color:blue;">Manual</mark>                 | Only when a support agent clicks the robot icon next to a customer message in the conversation | Works as originally configured; Copilot does not affect automatic replies |
| <mark style="color:blue;">Every incoming message</mark> | Automatically, for every customer message received                                             | No; the support agent decides whether to send the suggestion              |

{% hint style="warning" %}
After you choose "Every incoming message", the Agent no longer replies to customers automatically, and it no longer hands off automatically according to the [human handoff settings](https://docs.maiagent.ai/maiagent-user-guide/en/build/conversation-settings/human-handoff.md). Every customer message needs to be handled by a support agent, so make sure you have enough staff to reply promptly before switching to this mode.
{% endhint %}

## How Support Agents Use It <a href="#use-suggestions" id="use-suggestions"></a>

Open a conversation in <mark style="color:blue;">Customer Conversations</mark> → <mark style="color:blue;">All Conversations</mark>. On platforms with AI Copilot enabled, a robot icon appears below each customer message; hovering over it shows <mark style="color:blue;">Generate AI Suggestion</mark>.

<figure><img src="https://1360999650-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2F6v6TNkkOQVfRYfcNirHL%2Fuploads%2Fgit-blob-ee2f412f4312999a27932a7b4b29f762a9d64981%2Fcopilot-generate-icon.png?alt=media" alt="In a support conversation, the robot icon below a customer message; hovering over it shows Generate AI Suggestion"><figcaption><p>Robot icon below a customer message: click it to generate an AI suggestion</p></figcaption></figure>

1. **Generate a suggestion**: Click the robot icon, and an <mark style="color:blue;">AI Suggestion</mark> card appears below the message, with the content appearing word by word. In "Every incoming message" mode, suggestions are generated automatically without a click; if you want another version, you can still click the icon to generate one again.
2. **Check the content**: When the AI cites knowledge base content, the card also shows the sources so support agents can verify them. Click the card's title bar to collapse or expand the content.
3. **Send or edit**:
   * <mark style="color:blue;">Send</mark>: Sends the suggestion directly to the customer, with the same effect as a support agent sending a message from the input box.
   * <mark style="color:blue;">Edit</mark>: Puts the suggestion into the input box below so you can modify it and send it yourself. Any text already in the input box is replaced.
4. **Regenerate**: Click the refresh icon in the top-right corner of the card (<mark style="color:blue;">Regenerate</mark>), and the AI generates a new version.

<figure><img src="https://1360999650-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2F6v6TNkkOQVfRYfcNirHL%2Fuploads%2Fgit-blob-d8c2aea0752d0fcc0f95e2eb0d19a47bfc24d438%2Fcopilot-suggestion-versions.png?alt=media" alt="AI suggestion card: the title bar shows the ready status and a 2 / 2 version switcher, the top right has the regenerate and collapse icons, and the bottom right has the Edit and Send buttons"><figcaption><p>AI suggestion card: status "Ready", version switcher "2 / 2", the regenerate icon, and Edit and Send</p></figcaption></figure>

For how to send replies and the rules for taking over a conversation, see [Human Agent Takeover and Replies](https://docs.maiagent.ai/maiagent-user-guide/en/serve/daily-operations/human-handoff-operations.md).

### Switch Versions <a href="#switch-versions" id="switch-versions"></a>

When several suggestions have been generated for the same customer message, the card's title bar shows a version switcher, such as `2 / 3`. Use the left and right arrows (<mark style="color:blue;">Previous Version</mark>, <mark style="color:blue;">Next Version</mark>) to compare versions, then pick one to send or edit. While a new version is being generated, the card stays on the version being generated and you cannot switch for the time being.

### Suggestion Status <a href="#suggestion-status" id="suggestion-status"></a>

| Status                                      | Meaning                                                             | What you can do                                                                                                              |
| ------------------------------------------- | ------------------------------------------------------------------- | ---------------------------------------------------------------------------------------------------------------------------- |
| <mark style="color:blue;">Generating</mark> | The AI is generating the suggestion                                 | Wait for it to finish; you cannot send, edit, or regenerate during this time                                                 |
| <mark style="color:blue;">Ready</mark>      | The suggestion is complete                                          | <mark style="color:blue;">Send</mark>, <mark style="color:blue;">Edit</mark>, or <mark style="color:blue;">Regenerate</mark> |
| <mark style="color:blue;">Failed</mark>     | Generation did not succeed this time, and the card shows the reason | Click the regenerate icon to try again                                                                                       |

### Open the Copilot View in a New Tab <a href="#open-copilot-view" id="open-copilot-view"></a>

The robot icon in the conversation title bar (hovering over it shows <mark style="color:blue;">Conversation Copilot</mark>) opens a compact view of the conversation in a new tab, keeping only the messages, AI suggestions, and the input box, which makes it easy to use side by side with other systems.

## Credits Usage <a href="#credits" id="credits"></a>

Each AI suggestion generated consumes Credits just like an Agent reply, calculated based on the models and tools the Agent uses. Clicking <mark style="color:blue;">Regenerate</mark> produces a new suggestion and is billed again. Suggestions that fail to generate do not deduct Credits.

"Every incoming message" mode generates a suggestion for every customer message, so usage is usually higher than in Manual mode. For rates and consumption records, see [Credits Pricing](https://docs.maiagent.ai/maiagent-user-guide/en/others/credits.md#pricing-tab).

## FAQ <a href="#faq" id="faq"></a>

### I don't see the robot icon in the conversation. <a href="#faq-no-icon" id="faq-no-icon"></a>

Confirm that the platform this conversation belongs to has <mark style="color:blue;">Enable AI Copilot</mark> turned on in the <mark style="color:blue;">AI Copilot</mark> tab and that you saved. The robot icon appears only next to messages from customers, not next to messages sent by the AI or support agents.

### I clicked the robot icon and got "Suggestion generation failed." <a href="#faq-generate-failed" id="faq-generate-failed"></a>

Common causes are that AI Copilot has been turned off for the platform, or that the platform has no Agent bound. AI Copilot needs the platform's bound Agent to generate suggestions, so go to the platform's <mark style="color:blue;">Basic Settings</mark> and confirm that an Agent is selected.

### Will customers see AI suggestions? <a href="#faq-customer-visibility" id="faq-customer-visibility"></a>

No. AI suggestions appear only in the support agent's conversation view and are not sent to customer-facing channels such as websites or LINE. Customers receive content only when a support agent clicks <mark style="color:blue;">Send</mark>, or sends it themselves after editing.

### After switching to "Every incoming message", why has the AI stopped replying? <a href="#faq-no-auto-reply" id="faq-no-auto-reply"></a>

This is by design in this mode: the AI only generates suggestions, and the support agent decides whether to send them. To restore AI automatic replies, switch back to <mark style="color:blue;">Manual</mark> or turn off <mark style="color:blue;">Enable AI Copilot</mark>.


---

# 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 by asking a question.

Perform an HTTP GET request on the following URL with the `ask` and `goal` query parameters:

```
GET https://docs.maiagent.ai/maiagent-user-guide/en/serve/copilot.md?ask=<question>&goal=<user_goal>
```

`ask` is the immediate question: it should be specific, self-contained, and written in natural language.
`goal` is what the user is ultimately trying to achieve, the reason they need the answer. Sharing it helps GitBook give you a better, more relevant answer. A goal is most helpful when it describes the outcome the user wants rather than restating the question. For example, with `ask=how do I create an API token`, a goal like `automate deployments from our CI pipeline` lets GitBook tailor the answer to that use case.

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.
