> 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/build/agent/voice-agent.md).

# Voice Agent

Let Agents serve customers through "listening" and "speaking", supporting real-time conversations, STT/LLM/TTS pipelines, and interruption control

## What Is This? <a href="#what-is-this" id="what-is-this"></a>

Voice Agent enables an Agent to interact through **voice**: users speak, and the AI responds by speaking back. It integrates real-time voice models, STT (Speech-to-Text), TTS (Text-to-Speech), and interruption control under the hood, making it suitable for phone customer service, IVR, voice assistants, and similar scenarios.

Once enabled, a button for entering "voice interaction mode" appears in the input bar of MaiGPT, the Agent Marketplace, and Web Chat. Click it once to start talking. For usage, see [How Users Talk to the Agent](#how-users-talk).

## Three Interaction Modes <a href="#interaction-modes" id="interaction-modes"></a>

Different modes suit different needs, depending on latency, flexibility in customizing the voice, and tool support.

| Mode                | How It Works                                                    | Best For                                                                                                              |
| ------------------- | --------------------------------------------------------------- | --------------------------------------------------------------------------------------------------------------------- |
| **Realtime**        | Uses a voice model directly for real-time voice conversation    | Scenarios requiring the lowest latency where the default voice is acceptable                                          |
| **Realtime + TTS**  | A real-time voice model paired with custom TTS output           | Scenarios needing a custom brand voice while keeping a real-time feel                                                 |
| **STT + LLM + TTS** | The traditional pipeline: Speech-to-Text → LLM → Text-to-Speech | Scenarios requiring full tool support, where slightly higher latency is acceptable and maximum flexibility is desired |

**Quick decision**: For the fastest and most natural experience → Realtime; for a branded voice → Realtime + TTS; for maximum flexibility and tool support → STT + LLM + TTS.

## How to Fill In the Voice Agent Settings <a href="#voice-settings-walkthrough" id="voice-settings-walkthrough"></a>

<mark style="color:blue;">Voice Agent Settings</mark> are on the <mark style="color:blue;">Answer Mode Settings</mark> tab of the Agent settings page, after the Role Instructions section (they appear only if your organization has enabled AI Voice Agent). At the top is the "<mark style="color:blue;">Enable Voice Agent</mark>" master switch: when it is off, the whole settings group is collapsed, and the voice entry points in Web Chat, phone calls, and the back office will not offer voice service for this Agent. When it is on, fill in three sections in order; each collapsed section shows a summary of its current settings on the right.

### 1. Voice Technology <a href="#voice-settings-tech" id="voice-settings-tech"></a>

First, click a mode card under "<mark style="color:blue;">Voice Agent Mode</mark>" (click again to deselect), then choose providers and fill in configurations according to the mode:

| Mode            | Providers to Choose                                                                                                            | Configuration Fields                                                                                              |
| --------------- | ------------------------------------------------------------------------------------------------------------------------------ | ----------------------------------------------------------------------------------------------------------------- |
| Realtime        | <mark style="color:blue;">Realtime Voice Provider</mark>                                                                       | <mark style="color:blue;">Realtime Voice Configuration</mark>                                                     |
| Realtime + TTS  | <mark style="color:blue;">Realtime Voice Provider</mark>, <mark style="color:blue;">Text-to-Speech (TTS) Provider</mark>       | <mark style="color:blue;">Realtime Voice Configuration</mark>, <mark style="color:blue;">TTS Configuration</mark> |
| STT + LLM + TTS | <mark style="color:blue;">Speech-to-Text (STT) Provider</mark>, <mark style="color:blue;">Text-to-Speech (TTS) Provider</mark> | <mark style="color:blue;">STT Configuration</mark>, <mark style="color:blue;">TTS Configuration</mark>            |

Below each provider field is a side-by-side pair of fields: on the left, "<mark style="color:blue;">Configuration Example</mark>" is read-only; on the right is the JSON configuration that actually takes effect.

* Providers currently supported by the platform: for STT, OpenAI, Azure, and Bronci (Taiwanese Hokkien / Mandarin / Hakka); for TTS, OpenAI, Azure, VoAI, and Bronci; for realtime voice, OpenAI Realtime and Google Gemini Live. The provider list shows only providers enabled for your organization; if it is empty, "No providers available" is displayed — please contact MaiAgent to have them enabled.
* When you select a provider, its example configuration is automatically filled into the field on the right, and you can usually save as is; edit the JSON only if you want to adjust the voice, language, transcription model, and so on. The "<mark style="color:blue;">Configuration Reference</mark>" below the example lists each parameter's type, default value, allowed values, and description for you to follow, for example [User Transcript Settings](#user-transcript-settings).
* The configuration must be valid JSON; otherwise the field shows "Invalid configuration format. Please enter valid JSON", and saving will also expand the corresponding section so you can fix it. If the configuration is left empty, a reminder appears before saving: "The configuration is empty. The default configuration will be applied after saving." After you confirm, it is saved with the default configuration.

#### How to Choose a Realtime Voice Model <a href="#realtime-model-choice" id="realtime-model-choice"></a>

The options in "<mark style="color:blue;">Realtime Voice Provider</mark>" are displayed as "model name (provider)", for example `gemini-3.8-live (Google Gemini Live)` and `gpt-realtime-2.1 (OpenAI Realtime)`. Google Gemini Live currently offers two models:

| Model                                | Description                                                                                                                                                                                                                                                                                      |
| ------------------------------------ | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ |
| `gemini-3.8-live`                    | Listed first in the menu; it is the successor to `gemini-live-2.5-flash-native-audio`. Like 2.5, it can only be used in "<mark style="color:blue;">Realtime</mark>" mode, and "<mark style="color:blue;">Enable Voice Replies</mark>" must be on — see the note below                            |
| `gemini-live-2.5-flash-native-audio` | Existing model; Agents that already use it keep taking calls as usual. Google has announced this model's retirement (on Vertex AI, planned for December 13, 2026). Agents that already use it will not be switched to 3.8 automatically, so we recommend switching yourself before it is retired |

<figure><img src="https://1360999650-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2F6v6TNkkOQVfRYfcNirHL%2Fuploads%2Fgit-blob-a42c8ce063bb87b4f3fe44032e10ff60693090a4%2Fvoice-agent-realtime-model-options.png?alt=media" alt="Realtime Voice Provider dropdown; the first option is gemini-3.8-live (Google Gemini Live), followed by gemini-live-2.5-flash-native-audio and several OpenAI Realtime models"><figcaption><p>The "Realtime Voice Provider" menu: gemini-3.8-live is listed first, and 2.5 is still selectable</p></figcaption></figure>

For example, a customer service manager's phone-support Agent has always answered calls with `gemini-live-2.5-flash-native-audio` and wants to move to 3.8 before 2.5 is retired:

1. On the Agent settings page, under <mark style="color:blue;">Answer Mode Settings</mark> → <mark style="color:blue;">Voice Agent Settings</mark>, expand "<mark style="color:blue;">Voice Technology</mark>" and confirm that "<mark style="color:blue;">Voice Agent Mode</mark>" is "<mark style="color:blue;">Realtime</mark>".
2. In "<mark style="color:blue;">Realtime Voice Provider</mark>", select `gemini-3.8-live (Google Gemini Live)`. The "<mark style="color:blue;">Realtime Voice Configuration</mark>" on the right is replaced with the 3.8 example configuration, so any parameters you previously changed (for example the voice, `voice`) must be filled in again.
3. Click <mark style="color:blue;">Save</mark> at the bottom right, then open voice interaction mode and try a short conversation to confirm that you can hear the replies and that the conversation record has text on both sides.

{% hint style="warning" %}

* **Both Gemini Live models can only answer with voice**: `gemini-3.8-live` and `gemini-live-2.5-flash-native-audio` output only voice, not text, so they can only be used in **Realtime** mode, with "<mark style="color:blue;">Enable Voice Replies</mark>" turned on. In the following cases the voice model has to output text instead, Gemini Live refuses the connection, and the Agent will not respond during the call:

  * The Voice Agent mode is "<mark style="color:blue;">Realtime + TTS</mark>" (the voice model outputs text, which is then handed to TTS to be spoken).
  * "<mark style="color:blue;">Enable Voice Replies</mark>" in [Conversation Behavior](#voice-settings-behavior) is turned off (text-only replies).
  * The messaging platform that the call belongs to has [AI Copilot](/maiagent-user-guide/en/serve/copilot.md#trigger-modes) enabled with the trigger mode set to "<mark style="color:blue;">Every incoming message</mark>" (replies are decided by the support agent and not played directly).

  These combinations are not blocked when you save, so check them yourself before configuring. If you need any of the above, choose an OpenAI Realtime model for realtime voice, or use the "<mark style="color:blue;">STT + LLM + TTS</mark>" mode instead.
* **Billing**: Realtime voice calls consume Credits based on the voice and text input and output usage of the selected model. Rates differ between 3.8 and 2.5, so before switching you can compare them under <mark style="color:blue;">Organization Settings</mark> → <mark style="color:blue;">Credits</mark> → <mark style="color:blue;">Pricing</mark>. See [Credits Billing System](/maiagent-user-guide/en/others/credits.md).
* `gpt-realtime`, which OpenAI no longer offers, no longer appears in the menu. Agents that already use it keep taking calls and being billed as usual, but once you switch to another model and save, you cannot select it again.
  {% endhint %}

### 2. Realtime Voice Prompt <a href="#voice-settings-prompt" id="voice-settings-prompt"></a>

The "<mark style="color:blue;">Realtime Voice Prompt</mark>" is a prompt seen only by the voice model; if left empty, the Role Instructions are used. The hint above the field tells you which relationship is currently in effect, based on "voice mode × answer mode × whether it is filled in":

| On-Screen Hint                                                    | When                                                          | Meaning                                                                                                                                                                       |
| ----------------------------------------------------------------- | ------------------------------------------------------------- | ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| This mode does not support a voice prompt                         | STT + LLM + TTS                                               | This mode always uses the Role Instructions and the field cannot be edited; if you need a voice-specific prompt, use Realtime or Realtime + TTS                               |
| Currently in effect: using the Role Instructions                  | Realtime types × RAG Q\&A × empty                             | The voice model answers directly using the Role Instructions                                                                                                                  |
| Currently in effect: this prompt "replaces" the Role Instructions | Realtime types × RAG Q\&A × filled in                         | In RAG Q\&A, the knowledge base returns only retrieved snippets and the voice model generates the answer from this prompt, so the **Role Instructions have no effect at all** |
| Currently in effect: the voice layer uses the Role Instructions   | Realtime types × other answer modes such as Agent × empty     | The voice layer delegates the question back to the text reply flow, and the answer is still determined by the Role Instructions and tools                                     |
| Currently in effect: affects only the voice interaction layer     | Realtime types × other answer modes such as Agent × filled in | The answer content does not change; this prompt only adjusts how the voice model speaks (tone, manner, pauses)                                                                |

In RAG Q\&A mode, if you want to "fine-tune on top of the Role Instructions", click "<mark style="color:blue;">Use Role Instructions as Base</mark>" to copy the Role Instructions into the field and then edit them; if the field already has content, you will first be asked to confirm "Overwrite voice prompt".

### 3. Conversation Behavior <a href="#voice-settings-behavior" id="voice-settings-behavior"></a>

* "<mark style="color:blue;">Enable Voice Replies</mark>": When turned off, the Voice Agent replies with text only and does not play audio, which suits cases where you only want voice input.
* "<mark style="color:blue;">Interruption Control Settings</mark>": Allows users to interrupt the Voice Agent while it is responding. When on, you can adjust <mark style="color:blue;">Minimum Speech Duration (seconds)</mark> and <mark style="color:blue;">Minimum Word Count</mark>. These two fields take effect only in STT + LLM + TTS mode; in other modes a reminder is shown and the fields are disabled. See [Interruption Control](#turn-handling).

The "<mark style="color:blue;">Usage Notes</mark>" at the bottom of the section remind you that the Voice Agent searches all linked knowledge bases and cannot search specific files. After finishing the settings, click <mark style="color:blue;">Save</mark> at the bottom right.

## Interruption Control (Turn Handling) <a href="#turn-handling" id="turn-handling"></a>

A key part of the voice conversation experience — can the user interrupt the AI while it is speaking? How does the system tell whether the user is really speaking rather than making background noise or filler sounds?

Adjustable parameters:

| Parameter                             | Description                                                                              |
| ------------------------------------- | ---------------------------------------------------------------------------------------- |
| **Minimum speech duration (seconds)** | The user's speech must last at least this many seconds to count as "I want to interrupt" |
| **Minimum word count**                | The user must say at least this many words to trigger an interruption                    |

> **Note**: `min_duration` and `min_words` are only effective in **STT + LLM + TTS** mode. In Realtime mode, interruption detection is handled internally by the voice model.

## User Transcript Settings (Realtime Mode) <a href="#user-transcript-settings" id="user-transcript-settings"></a>

In Realtime and Realtime + TTS modes, the voice model listens to the user's voice directly, and the text record of what the user said (the user transcript) is produced by a separate transcription model. When <mark style="color:blue;">Realtime Voice Provider</mark> is set to OpenAI, you can use `input_audio_transcription` in the JSON of <mark style="color:blue;">Realtime Voice Configuration</mark> to specify how transcription is done:

```json
{
  "input_audio_transcription": {
    "model": "gpt-4o-mini-transcribe",
    "language": "zh",
    "prompt": "Output in Traditional Chinese. Common terms: MaiAgent, insurance policy, claims"
  }
}
```

| Field      | Description                                                                                                                                                             |
| ---------- | ----------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| `model`    | The transcription model; can be `gpt-4o-transcribe`, `gpt-4o-mini-transcribe`, `gpt-transcribe`, or `gpt-live-transcribe`; if not set, `gpt-4o-mini-transcribe` is used |
| `language` | A language hint as an ISO-639-1 code (for example `zh`, `en`, `ja`); if left empty, the model detects it automatically                                                  |
| `prompt`   | A hint for the transcription model, such as proper nouns you expect to appear, or an instruction to output in Traditional Chinese                                       |

The settings page lists each field's type, default value, and allowed values below the configuration field for you to follow.

* **Affects only the transcript**: These settings determine the text on the user's side of the call record; they do not change what the voice model hears or how it answers.
* **Multilingual calls**: If users may mix several languages, leave `language` empty so the model detects it automatically; if the language is always a single one, fill in its code for more stable recognition.
* **Billing**: The user transcript is charged to Credits by seconds of speech-to-text, according to the transcription model actually used. For each model's unit price, see <mark style="color:blue;">Organization Settings</mark> → <mark style="color:blue;">Credits</mark> → <mark style="color:blue;">Pricing</mark>; for details, see [Credits Billing System](/maiagent-user-guide/en/others/credits.md).

## Conversation States <a href="#conversation-states" id="conversation-states"></a>

While a voice conversation is in progress, the interface shows the current state:

* **Listening**: The AI is receiving the user's voice
* **Thinking**: The AI is processing (searching the knowledge base, using tools, generating a reply)
* **Replying**: The AI is replying by voice
* **Initializing**: The connection has just been established and is getting ready

## How Users Talk to the Agent <a href="#how-users-talk" id="how-users-talk"></a>

Once an Agent has the Voice Agent enabled, users need no setup and can switch to voice from the usual text conversation screen:

| Where the User Is                                                                                                                      | Entry Point                                                                                                                                                                                                                                                                                                                                                                                                                        |
| -------------------------------------------------------------------------------------------------------------------------------------- | ---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| The conversation page of the back-office <mark style="color:blue;">MaiGPT</mark> or <mark style="color:blue;">Agent Marketplace</mark> | An extra <mark style="color:blue;">Voice Interaction Mode</mark> button appears on the right side of the input bar. The organization must have AI Voice Agent enabled, and the Agent in the current conversation must have the Voice Agent enabled                                                                                                                                                                                 |
| Web Chat on a website                                                                                                                  | A <mark style="color:blue;">headset</mark> icon next to the input box (different from the microphone icon for [voice input](/maiagent-user-guide/en/build/choose-channel/web-chat.md#voice-input-and-playback): the microphone speaks one sentence and sends it as text; the headset is a live call)                                                                                                                               |
| A standalone voice conversation page                                                                                                   | Next to the messaging platform's <mark style="color:blue;">Public Access URL</mark> is a <mark style="color:blue;">Voice Mode</mark> link (see [Basic Settings](/maiagent-user-guide/en/build/conversation-settings/platform-settings/basic-settings.md#public-url-and-embed)); it opens a page that only has the call: click <mark style="color:blue;">Start Conversation</mark>, and after "Dialing" is shown, the call connects |

### Voice Interaction Mode Window <a href="#voice-mode-window" id="voice-mode-window"></a>

After you click the button, the browser first asks for microphone permission (the first time you use it), the button shows "Connecting...", and then the <mark style="color:blue;">Voice Interaction Mode</mark> window opens:

| Area                                                                                                  | Description                                                                                                                                                                                                                                                                                                                        |
| ----------------------------------------------------------------------------------------------------- | ---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| <mark style="color:blue;">Usage Notes</mark>                                                          | Reminds you that Realtime and Realtime + TTS modes support only MCP tools, and that the Voice Agent searches all linked knowledge bases (see [Usage Limitations](#limitations))                                                                                                                                                    |
| <mark style="color:blue;">Connection Status</mark>                                                    | Not connected → Connecting... → Connected; if it fails, "Connection failed" is shown                                                                                                                                                                                                                                               |
| Current state                                                                                         | <mark style="color:blue;">Initializing</mark> / <mark style="color:blue;">Listening</mark> / <mark style="color:blue;">Thinking</mark> / <mark style="color:blue;">Replying</mark>; for their meanings, see [Conversation States](#conversation-states). When the Agent delegates to another Agent, "Sub-agent replying…" is shown |
| <mark style="color:blue;">Conversation Record</mark>                                                  | A live text record of this call; before the call connects, "Waiting for the conversation to start..." is shown                                                                                                                                                                                                                     |
| <mark style="color:blue;">Mute Microphone</mark> / <mark style="color:blue;">Unmute Microphone</mark> | Temporarily mute or resume; while recording, a "Recording" indicator is shown                                                                                                                                                                                                                                                      |
| <mark style="color:blue;">End Call</mark>                                                             | Disconnects and closes the window. The call content stays in the original conversation as text, and you can continue by typing                                                                                                                                                                                                     |

### Common Messages <a href="#voice-mode-messages" id="voice-mode-messages"></a>

| On-Screen Text                                                                              | What to Do                                                                                                                                                                         |
| ------------------------------------------------------------------------------------------- | ---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| "Microphone permission denied" / "Please allow microphone permission to use voice features" | Allow the microphone in the site permissions to the left of the browser address bar, then click again                                                                              |
| "No microphone device found"                                                                | Make sure a microphone or headset is connected to the computer and selected in the system sound settings                                                                           |
| "Assistant connection timed out. Please try again later"                                    | The voice service is temporarily not connected; wait a few seconds and click again. If it keeps happening, contact your administrator to check the Agent's voice provider settings |
| "Unable to start voice interaction mode"                                                    | The system cannot establish a voice connection for the moment; try again later. If it keeps happening, contact your administrator                                                  |

## How Is It Different from Regular Text Conversations? <a href="#voice-vs-text" id="voice-vs-text"></a>

|                         | Text Agent                                          | Voice Agent                                                        |
| ----------------------- | --------------------------------------------------- | ------------------------------------------------------------------ |
| **Input**               | Typing on a keyboard, pasting files                 | Microphone voice, possibly with DTMF key presses                   |
| **Output**              | Text, Markdown, images, files                       | Voice, with a text transcript attached at the end                  |
| **Latency requirement** | Seconds are acceptable                              | Must be at the millisecond level to feel natural                   |
| **Best for**            | Detailed information, records you will look back on | Immediate responses, hands-busy situations, phone customer service |

## When to Use It <a href="#when-to-use" id="when-to-use"></a>

* **Phone customer service**: Replace traditional IVR; the AI answers the phone, listens to the question, and gives the answer
* **Voice query systems**: Customers call to check orders, balances, or insurance policies
* **Voice FAQ**: Ask common questions directly by speaking
* **Driving / hands-busy situations**: Users cannot type but need AI assistance
* **Accessibility needs**: Friendlier to users who have difficulty typing

## Usage Limitations <a href="#limitations" id="limitations"></a>

* **Tool support**: Currently, **Realtime** and **Realtime + TTS** modes support only **MCP tools**; to use API tools, switch to STT + LLM + TTS mode
* **Knowledge base search**: The Voice Agent searches all linked knowledge bases and **cannot be limited to searching only certain documents**
* **Gemini Live models**: `gemini-3.8-live` and `gemini-live-2.5-flash-native-audio` can only be used in **Realtime** mode, with "<mark style="color:blue;">Enable Voice Replies</mark>" turned on; if combined with Realtime + TTS, with voice replies turned off (text-only replies), or used on an AI Copilot messaging platform whose trigger mode is "Every incoming message", the call cannot respond properly. See [How to Choose a Realtime Voice Model](#realtime-model-choice)
* **Microphone permission**: The user must authorize the browser to use the microphone before a conversation can start

## What Do I Need to Do? <a href="#what-do-i-need-to-do" id="what-do-i-need-to-do"></a>

1. **Open the Agent settings** — On the settings page of the Agent you want to enable voice for, switch to <mark style="color:blue;">Answer Mode Settings</mark>, find <mark style="color:blue;">Voice Agent Settings</mark>, and turn on <mark style="color:blue;">Enable Voice Agent</mark>. This section appears only if your organization has enabled AI Voice Agent
2. **Choose the Voice Agent mode** — Pick one of Realtime / Realtime + TTS / STT + LLM + TTS
3. **Set the providers and configuration** — Choose the matching STT / TTS / Realtime providers for the mode; the example configuration is filled in automatically once selected. See [Voice Technology](#voice-settings-tech)
4. **Decide on a voice prompt** (optional) — Leave it empty to use the Role Instructions, or fill it in following the table in [Realtime Voice Prompt](#voice-settings-prompt)
5. **Adjust conversation behavior** (optional) — Whether to play voice replies and whether to allow interruptions; in STT + LLM + TTS mode you can adjust the minimum duration and minimum word count
6. **Test** — Open a voice call in the interface and confirm that the connecting, listening, and replying states all run smoothly

## Further Reading <a href="#further-reading" id="further-reading"></a>

* [Voice Customer Service: From Zero to Launch](/maiagent-user-guide/en/serve/voice.md)
* [Phone Customer Service (Call Center)](/maiagent-user-guide/en/build/choose-channel/call-center.md)
* [Voice Customer Service Scenarios Overview](/maiagent-user-guide/en/serve/voicecs.md)
* [IVR Intent Recognition](/maiagent-user-guide/en/serve/voicecs/ivr-intent-recognition.md)
* [Voice Call Summary](/maiagent-user-guide/en/serve/voicecs/call-summary.md)
* [Voice Call Quality Inspection](/maiagent-user-guide/en/serve/voicecs/call-qa.md)


---

# 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/build/agent/voice-agent.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.