> ## Documentation Index
> Fetch the complete documentation index at: https://docs.ambivertai.tech/llms.txt
> Use this file to discover all available pages before exploring further.

# Asterisk ARI Integration

> Connect Ambivert AI AI to your Asterisk PBX using the Asterisk REST Interface (ARI)

## Overview

Asterisk ARI (Asterisk REST Interface) allows you to connect Ambivert AI AI voice agents to your existing Asterisk PBX. ARI provides a WebSocket-based event model for controlling calls via Stasis applications, giving Ambivert AI full control over call flow and audio streaming.

This guide focuses on the Ambivert AI-specific configuration. For general Asterisk installation and administration, refer to the [official Asterisk documentation](https://docs.asterisk.org/).

## Prerequisites

Before setting up the ARI integration, ensure you have:

* A running Asterisk instance with `chan_websocket` and `res_websocket_client` modules available. Known-working setups: (a) Asterisk 22+, (b) Asterisk 20 LTS with these modules included
* ARI module enabled in Asterisk
* `chan_websocket` (WebSocket channel driver) and `res_websocket_client` (loads `websocket_client.conf`) enabled in your Asterisk build. Verify with `asterisk -rx "module show like chan_websocket"` and `asterisk -rx "module show like res_websocket_client"` — both should report **Running**.
* Network connectivity between your Ambivert AI instance and Asterisk
* Ambivert AI AI instance running and accessible

<Note>
  If you compiled Asterisk from source, ensure both `chan_websocket` and `res_websocket_client` are included during the build. These modules are required for external media streaming between Asterisk and Ambivert AI. Refer to the [Asterisk build system documentation](https://docs.asterisk.org/) for details on enabling modules.
</Note>

## Asterisk Configuration

The following Asterisk configuration files need to be set up to work with Ambivert AI. These are minimal examples focused on the Ambivert AI integration -- refer to the [Asterisk documentation](https://docs.asterisk.org/) for full configuration details.

### Enable ARI (`ari.conf`)

Create an ARI user that Ambivert AI will use to authenticate:

```ini theme={null}
[general]
enabled = yes

[ambivert]
type = user
read_only = no
password = your_secure_password
```

<Note>
  The username (section name, e.g., `ambivert`) and password here must match the **Stasis App Name** and **App Password** you configure in Ambivert AI.
</Note>

### Enable the HTTP Server (`http.conf`)

ARI requires the Asterisk HTTP server to be enabled:

```ini theme={null}
[general]
enabled = yes
bindaddr = 0.0.0.0
bindport = 8088
```

### Configure the Stasis Dialplan (`extensions.conf`)

Route incoming calls to your Stasis application so Ambivert AI can handle them:

```ini theme={null}
[from-external]
exten => _X.,1,NoOp(Incoming call to ${EXTEN})
 same => n,Stasis(ambivert)
 same => n,Hangup()
```

Replace `ambivert` with the app name you configured in `ari.conf` and in Ambivert AI.

### Configure External Media Streaming (`websocket_client.conf`)

Ambivert AI uses Asterisk's external media streaming to send and receive audio over WebSocket. Configure a WebSocket client connection that points to your Ambivert AI instance:

<Tabs>
  <Tab title="Ambivert AI Cloud">
    ```ini theme={null}
    [ambivert]
    type = websocket_client
    uri = wss://api.ambivertai.tech/api/v1/telephony/ws/ari
    protocols = media
    tls_enabled = yes
    ca_list_file = /etc/ssl/certs/ca-certificates.crt
    ```

    <Note>
      `tls_enabled = yes` is required even though the URI scheme is `wss://` — without it Asterisk will not negotiate TLS and the connection will fail. The ARI credentials (**Stasis App Name** and **App Password**) must match what you configure in the Ambivert AI dashboard under Telephony Settings.
    </Note>
  </Tab>

  <Tab title="Self-hosted">
    ```ini theme={null}
    [ambivert]
    type = websocket_client
    uri = ws://your-ambivert-host:port/api/v1/telephony/ws/ari
    protocols = media
    ```

    <Note>
      Self-hosted deployments on an internal network may use an unencrypted WebSocket (`ws://`). If your Ambivert AI instance is exposed over HTTPS, use `wss://` and the corresponding hostname instead.
    </Note>
  </Tab>
</Tabs>

<Note>
  The section name (e.g., `ambivert`) is the **WebSocket Client Name** you'll enter in the Ambivert AI telephony configuration. This name tells Asterisk which WebSocket connection to use for external media streaming during calls.
</Note>

<Note>
  Ambivert AI's external media channel uses **G.711 μ-law (`ulaw`)**. Make sure any PJSIP endpoint or SIP trunk that places or receives calls through Ambivert AI allows `ulaw` (e.g. `allow=ulaw` in the endpoint config).
</Note>

Refer to the [Asterisk WebSocket documentation](https://docs.asterisk.org/) for additional `websocket_client.conf` options and TLS configuration.

### Apply the configuration changes

After editing any of the files above, reload the affected Asterisk modules from the Asterisk CLI (`asterisk -rvvv`):

```bash theme={null}
ari reload                                 # picks up ari.conf changes
dialplan reload                            # picks up extensions.conf changes
module reload res_websocket_client.so      # picks up websocket_client.conf changes
```

Changes to `http.conf` require a full Asterisk reload (`core reload`) or a service restart.

## Configuration in Ambivert AI

### Step 1: Navigate to Telephony Settings

1. Navigate to **/telephony-configurations** and click **Add configuration**
2. Select **Asterisk ARI** as your provider

### Step 2: Enter Your ARI Credentials

Configure the following fields:

| Field                     | Description                                                 | Example                            |
| ------------------------- | ----------------------------------------------------------- | ---------------------------------- |
| **ARI Endpoint URL**      | HTTP base URL of your Asterisk ARI server                   | `http://asterisk.example.com:8088` |
| **Stasis App Name**       | The ARI username configured in `ari.conf`                   | `ambivert`                         |
| **App Password**          | The ARI password configured in `ari.conf`                   | `your_secure_password`             |
| **WebSocket Client Name** | The connection name from `websocket_client.conf`            | `ambivert`                         |
| **From Extensions**       | Optional SIP extensions or trunk numbers for outbound calls | `PJSIP/6001` or `6001`             |

### Step 3: Save and Add Extensions

1. Click **Save Configuration**
2. Open the configuration you just created and add each SIP extension that should be reachable as a **phone number** (e.g. `8000`). For inbound, you'll assign a workflow to each extension separately — see [Inbound Calling](#inbound-calling) below.
3. Create a test workflow and initiate a test call to verify the connection.

## Inbound Calling

Unlike other telephony providers that use HTTP webhooks for inbound calls, ARI delivers inbound calls as **StasisStart events on the ARI WebSocket**. Ambivert AI automatically detects these events and activates the workflow assigned to the called extension.

### How It Works

1. An external call arrives at Asterisk and the dialplan routes it to `Stasis(ambivert)`
2. Asterisk fires a StasisStart event over the ARI WebSocket with the channel in `Ring` state and the dialed extension in the dialplan context
3. Ambivert AI looks up the called extension in your telephony configuration's phone numbers, finds the assigned workflow, validates quota, and creates a workflow run
4. The call is answered, bridged to an external media channel, and your voice agent workflow begins

Workflow assignment is **per extension**, so different extensions on the same Asterisk can route to different agents.

### Setting Up Inbound Calls

**Step 1: Configure the Asterisk dialplan**

Ensure your dialplan routes the extensions you care about into the Stasis application. Either route a specific extension:

```ini theme={null}
[from-external]
exten => 8000,1,NoOp(Incoming call to 8000)
 same => n,Stasis(ambivert)
 same => n,Hangup()
```

…or use a pattern that catches every extension you'll register in Ambivert AI:

```ini theme={null}
[from-external]
exten => _X.,1,NoOp(Incoming call to ${EXTEN})
 same => n,Stasis(ambivert)
 same => n,Hangup()
```

Replace `ambivert` with the app name you configured in `ari.conf` and in Ambivert AI.

**Step 2: Add the extension as a phone number in Ambivert AI**

1. Go to **/telephony-configurations** and open your Asterisk ARI configuration
2. In the **Phone numbers** section, add a phone number whose address is the SIP extension (e.g. `8000`)
3. Set its **Inbound workflow** to the agent that should answer
4. Save

   <Note>
     Adding the extension in Ambivert AI doesn't change Asterisk's dialplan — that's
     what Step 1 is for. The Ambivert AI entry tells the StasisStart handler which
     workflow to run when a call to that extension reaches the Stasis app.
   </Note>

Repeat Step 2 for each extension that should reach a voice agent.

**Step 3: Test an inbound call**

Place a call to one of the extensions you configured. You should see the assigned workflow activate and the voice agent respond.

### Inbound Call Context

When an inbound call activates a workflow, the following context is available to your workflow:

| Field           | Description                            |
| --------------- | -------------------------------------- |
| `caller_number` | The caller's phone number or extension |
| `called_number` | The dialed number or extension         |
| `direction`     | Always `inbound`                       |
| `call_id`       | The Asterisk channel ID                |
| `provider`      | Always `ari`                           |

## Troubleshooting

<AccordionGroup>
  <Accordion title="Cannot connect to ARI endpoint">
    * Verify the ARI endpoint URL is correct and reachable from your Ambivert AI instance
    * Check that the Asterisk HTTP server is running (`http.conf` has `enabled = yes`)
    * Ensure firewall rules allow traffic on the ARI port (default: 8088)
    * Confirm the ARI module is loaded: run `module show like res_ari` in the Asterisk CLI
  </Accordion>

  <Accordion title="Authentication failed">
    * Verify the Stasis App Name matches the ARI user section name in `ari.conf`
    * Check the App Password matches the password in `ari.conf`
    * Ensure there are no extra spaces in the credentials
  </Accordion>

  <Accordion title="No audio during calls">
    * Verify `chan_websocket` is loaded: run `module show like chan_websocket` in the Asterisk CLI
    * Check that `websocket_client.conf` is correctly configured with the right Ambivert AI URI
    * Ensure the WebSocket Client Name in Ambivert AI matches the section name in `websocket_client.conf`
    * Verify network connectivity and firewall rules allow WebSocket traffic between Asterisk and Ambivert AI
  </Accordion>

  <Accordion title="Calls not reaching Ambivert AI">
    * Ensure the dialplan routes calls to `Stasis(your_app_name)`
    * Verify the app name in the dialplan matches the ARI user in `ari.conf`
    * Check Asterisk CLI for errors: `asterisk -rvvv`
    * Confirm the ARI WebSocket connection is active
  </Accordion>

  <Accordion title="Inbound calls are immediately hung up">
    * Verify the called extension is added as a phone number under your ARI
      configuration in /telephony-configurations and has an **Inbound workflow**
      assigned
    * Confirm the workflow exists and belongs to the same organization as the
      ARI config
    * Check that your organization has available quota
    * Review Ambivert AI logs for warnings like "no matching phone number registered
      for config" or "has no inbound\_workflow\_id assigned"
  </Accordion>

  <Accordion title="WebSocket client connection issues">
    * Check the URI in `websocket_client.conf` points to the correct Ambivert AI host and port
    * Verify the Ambivert AI instance is running and accepting WebSocket connections
    * If using TLS, ensure certificates are correctly configured on both sides
  </Accordion>
</AccordionGroup>

## Best Practices

* Keep your Asterisk instance on the same network or a low-latency connection to Ambivert AI for optimal audio quality
* Use strong passwords for ARI authentication
* Restrict ARI access to known IP addresses using firewall rules
* Monitor Asterisk logs alongside Ambivert AI logs when debugging call issues
* Keep Asterisk updated to the latest stable version for security and compatibility

## Further Reading

* [Asterisk Documentation](https://docs.asterisk.org/) -- official reference for all Asterisk configuration
* [ARI Documentation](https://docs.asterisk.org/Configuration/Interfaces/Asterisk-REST-Interface-ARI/) -- detailed ARI configuration and API reference