Send Outbound Messages via SMS, WhatsApp and Other Channels

Flex allows your agents to initiate conversations with your customers via any Twilio Messaging Channel, including SMS, WhatsApp, and Facebook Messenger.

Currently, outbound messages require additional development work. This page outlines some of the work you'll need to do for the backend but outbound messages will also require a Flex UI Plugin to expose the functionality in the Flex interface.

Managing Task Creation

There are two ways to handle task creation for outbound messages: when the agent messages the customer (immediately) or when the customer responds to the outbound message (delayed).

Strategy 1: Create Task When Agent Messages Customer

The following steps will create a task when you send an outbound message to your customer.

1. Define a Flex Flow

First, define an outbound Flex Flow with an Integration type of task. You only need to do this once. Please ensure you are also defining your Flex Flow to use a Messaging capable Task Channel such as sms or chat. If you already have a Flex Flow that's using Studio, then the enabled parameter must be set to false.

2. Create a Flex Chat Channel

Next, create a Chat Channel using the Flex API to initiate messaging with the customer. In this example, the API call also adds Task Attributes, which will allow the channel to route the task to a specific agent (using the targetWorker Workflow expression).

Make sure to save the SID of the Chat Channel - you'll need it later!

Once the Flex Channel API is executed succesfully, two things happen:

• A new chat channel is created
• A new task is created and assigned to the worker specified in targetWorker

Then, if you open the Flex UI for targetWorker, you should see the following new reservation:

3. Set up a Twilio Proxy Session

Now that you have a channel, you'll want to associate it with a Proxy Session. Sessions offer a useful layer of state and metadata that Flex uses to ensure that your customers' messages stay properly threaded as they are processed by bots, TaskRouter and agents.

You'll need to make two requests:

The first creates the Proxy Session, and includes the person you're trying to reach.

The second request adds the chat channel to the session, and maps it to the Contact Center number.

In the sample code, you need to replace:

You'll also need to add the agent to the Proxy Session, using their Chat Channel SID as their identifier and the Contact Center number as their Proxy identifier, since messages from the agent should look like they're coming from the Contact Center.

Finally, if you want the agent to see the Customer's number when they respond to the chat, make the Agent's friendly name the customer's number.

In the sample code, you need to replace:

3. Update the Chat Channel Attributes

Update the Chat Channel attributes to include the Proxy Session SID you created in Step 3 if it's not already there. The chat channel attributes should be a JSON string:

{proxySession: "KCXXXXXXXXXXXXXXXXXXXXXXXX"}

Make sure that you retrieve the existing attributes object, add the proxySession key, and then update the parameter as a string.

Node.js Example:

.then(
    attributes => {
        return Object.assign(JSON.parse(attributes.attributes), { proxySession: proxySession.sid })
    }
))

4. Start chatting

Now, go back to Flex and accept the reservation. Your agent can directly message into the Chat interface, and messages are routed to the customer on their Twilio Channel.

Strategy 2: Create Task When Customer Responds

The following steps outline an alternate architecture for sending a message to the customer. When the customer responds, the inbound message handler will create a Task and forward it to an agent.

1. Configure inbound messages

Make sure you set up an inbound message handler, and that it's configured to create a Task. This will handle the customer's response to the outbound message. Please ensure you are also defining your Flex Flow to use a Messaging capable Task Channel such as sms or chat.

If you already handle inbound messages and want to use that logic, you can use the code sample below to retrieve all Flex Flows on the account - simply choose the SID of the one associated with your current message handler for your desired channel (e.g., the SMS message handler).

2. Create a Channel

Create a chat channel using the Flex Channel Resource. You'll need the Flex Flow for outbound messages. You'll need a new channel for each interaction.

3. Create a Proxy Session

Now you'll need to create a Proxy Session in order to associate the Chat Channel and the Customer's Phone number together. Sessions offer a useful layer of state and metadata that Flex uses to ensure that your customers' messages stay properly threaded as they are processed by bots, TaskRouter and agents. It will help tell Flex that, when the customer responds, they should be talking to this Agent over this Channel. Pass the channel you created in Step 2 as the Unique Name of the Session.

Make sure that you inclued the 'FriendlyName' in the participants array.

In the sample code, you need to replace:

Use another request to add the customer, using their number as the identifier. Proxy can use its own logic to select the Proxy identifier.

4. Update the Chat Channel Attributes

Update the Chat Channel attributes to include the Proxy Session SID you created in Step 3 if it's not already there. The chat channel attributes should be a JSON string:

{proxySession: "KCXXXXXXXXXXXXXXXXXXXXXXXX"}

Make sure that you retrieve the existing attributes object, add the proxySession key, and then update the parameter as a string.

5. Send the message

Now you can manually send your outbound message through the chat channel. This will trigger a post-event webhook, creating the proxy interaction or the outbound message. For example, if you're using Studio, then it will trigger the Studio Flow as an outbound message to your customer.

We advise you to send the message through the channel SID as opposed to the Proxy SID to keep a full record of the chat interaction.

Send the message to the chat channel

const sendMessage = async function(channelSid, chatServiceSid, client, from, messageText){
  try { 
    console.log("Send Message Function")
    return client
      .chat
      .services(chatServiceSid)
      .channels(channelSid)
      .messages
      .create({
        body : messageText,
        from : from
      })
  } catch(error) {
    console.log(error);
  }
}

Your customer should now receive a message. When they respond, a Task will be created, and you can route it to an agent who can accept and handle the response.