Zendesk Messaging

Overview

Zendesk Messaging Glass is Ada’s integration with the Zendesk Messaging asynchronous chat product. This integration enables Handoffs from your AI Agent to live agents in Zendesk, giving you access to Ada’s full feature set within your Zendesk environment.

This topic is about Zendesk Messaging. If you’re looking for information about Zendesk Chat, please see Zendesk Chat Handoff usage.

What is asynchronous chat?

Asynchronous chat means that conversations between end users and your live support agents can extend indefinitely, until either the end user or agent ends the chat manually. It is not limited by automatic timeouts or disconnects, as is the case with session-based chat platforms like Zendesk Chat.

Requirements

Ada Glass for Zendesk Messaging requires that your Zendesk and Ada accounts have the following products enabled and configured:

If you’re migrating from Zendesk Chat to Zendesk Messaging, inform your Customer Success Consultant (CSC). Specific configuration changes must be made during the migration process to ensure a smooth transition.

Capabilities & configuration

Zendesk Messaging Glass provides Handoff capabilities for asynchronous conversations between end users and live agents.

Key capabilities

  • Asynchronous conversations: Conversations can extend indefinitely without automatic timeouts.
  • Seamless transitions: End users and agents interact in the same chat window.
  • Persistent history: Conversation history persists across sessions unless manually cleared.
  • Department routing: Route Handoffs to specific Zendesk departments.
  • Zendesk field mapping: Map Ada variables to Zendesk ticket fields.
  • Automatic return: End users automatically return to your AI Agent when tickets are solved.

Quick start

Complete these steps to set up Zendesk Messaging Glass for Handoffs. See Implementation & usage for detailed instructions.

1

Configure Zendesk Messaging Glass

Enter your Zendesk subdomain, login email, and API token in the Ada dashboard.

2

Create an end conversation trigger

Set up a Zendesk trigger to automatically end agent conversations when tickets are solved.

3

Configure conversation control

Set your Zendesk Messaging conversation control to Pass control.

4

Connect your AI Agent

Add your AI Agent in Zendesk Admin Center as the first responder.

5

Add the Zendesk Messaging block

Add and configure the Zendesk Messaging block in your Handoff dialog.

Implementation & usage

Configure Zendesk Messaging Glass and set up the Handoff block by completing the following steps.

Configure Zendesk Messaging Glass

Connect Ada to your Zendesk Messaging environment.

You should have your Sunshine Conversations app open while configuring Zendesk Messaging Glass. You’ll need it to retrieve the required authentication information needed to complete the set up.

To configure Zendesk Messaging Glass:

  1. On the Ada dashboard, go to Config > AI AGENT > Handoffs. Then, on the Integrations tab, beside Zendesk Messaging, click Configure.

    The Connect Ada to Zendesk Messaging window opens.

  2. In the Connect Ada to Zendesk Messaging window, click Next.

  3. In the Configure Zendesk Messaging dialog box, complete the following fields using your Zendesk credentials, found under Settings in your Sunshine Conversations app.

    • Zendesk Subdomain

    • Zendesk Login Email

    • Zendesk API Token

      Refer to the Zendesk user guide Generating a new API token for guidance on creating an API token for your Zendesk Messaging integration.

  4. Click Finish, then close the success dialog box.

Create an end conversation trigger

Create a trigger in Zendesk to automatically end the agent conversation once the ticket status is set to solved. This trigger releases the end user from the Zendesk agent conversation and allows them to resume chatting with your AI Agent.

Zendesk may change their functionality without notice. For more details and the most up to date information about creating triggers, refer to Zendesk’s guide Creating triggers for automatic ticket updates and notifications.

To create an end conversation trigger:

  1. In the Zendesk Admin Center, go to Objects and rules > Business rules > Triggers, then click Create trigger.

  2. On the Add new trigger page, add a trigger name, description, and category.

  3. Under Meet ALL of the following conditions, click Add condition, then use the drop-down menus to set the condition to:

    • Category: Ticket > Status
    • Operator: Is
    • Value: Solved
  4. Under Actions, click Add action, then use the drop-down menus to set the action to:

    • Category: Ticket > Status
    • Value: Closed
  5. Click Create trigger.

With the trigger in place, end users are now automatically released back to your AI Agent when the agent selects Submit as Solved at the end of a live support conversation.

End users on web or mobile can release themselves from an agent conversation by clicking the X at the top of the chat window.

Let your AI Agent pass the conversation back to Ada

After the live agent has resolved the end user’s inquiry, Zendesk Messenger can pass the conversation back to Ada.

To configure conversation control:

In your Zendesk Messaging settings, ensure that your Conversation control setting is set to Pass control. For information on how to do this, see Configuring conversational control at Zendesk’s documentation.

Connect your AI Agent with Zendesk Admin Center

In Zendesk, add your AI Agent so it can be the first means of interacting with end users. For more information, see Managing third-party bots in Admin Center at Zendesk’s documentation.

Here’s a summary of what the actions do:

  • Connect - Makes a switchboard setup where Ada is the first responder, Zendesk Agent Workspace is the second, and Zendesk’s Answer Bot is not connected to any other switchboard integrations.

  • Disconnect - Sets Answer Bot back as the first responder, but keeps Ada in the switchboard in case you want to reconnect it in the future. This can be handy while you’re testing your AI Agent.

  • Uninstall - Removes Ada from your switchboard, and uninstalls it from your Sunshine Conversations app. This is not recommended; if you do, you’ll have to reinstall Ada in your Zendesk account again.

If you can’t see the above actions, you can perform the same actions in the Sunshine Conversations API. For instructions on how to do that, see Configuring the switchboard with Zendesk and a bot partner at Sunshine Conversations’ documentation. If you continue to have difficulty, contact your Ada team for assistance.

Sunshine Conversations Web Messenger won’t appear in your list of channels in Zendesk Admin Center. However, it’s accessible in the Sunshine Conversations API if you need to see or update it.

Use the Zendesk Messaging block

Add the Zendesk Messaging block to your Handoff dialog to route end users to live agents.

To add and configure the Zendesk Messaging block:

  1. On the Ada dashboard, go to Config > AI AGENT > Handoffs, then create or select a Handoff to edit. For more information, see Manage Handoffs.

  2. Optional. In the Handoff content, add Capture blocks configured to capture the end user’s name, email address, and phone number.

    In order for the end user’s information to map to the correct fields on the Zendesk ticket, the variables you use to capture their name and email address must be named exactly as they appear below:

    • name
    • email
  3. Drag and drop the Zendesk Messaging block into the Handoff content, then configure the block using the following table as a guide:

    Standby Message

    Support requests can occur both during business hours, or during off hours. Use this message to acknowledge the end user’s request and present information about response times.

    Craft a message for end users with helpful information, such as average wait times, or if agents are offline. Use the Scheduled block to help make this an even smoother experience.

    Department

    This drop-down menu lists all the available support departments. Select the appropriate department for this Handoff.

    Zendesk Fields

    Map your Zendesk fields to the block. Use these fields to enhance your support process:

    • Present your agents with key end user details.

    • Use Support triggers to route incoming tickets to the right team.

    You can map standard Zendesk fields such as Requester, Follower, Assignee, CCs, Share, Status, Type, Priority and Tags. These fields help ensure tickets are created with the right context and routed appropriately from the start.

    Zendesk Field types Text, Number, Boolean, and Dropdown field types are all accepted.

    • If you’re using the Dropdown field type, make sure you send the Tag value associated with the dropdown to Zendesk, to ensure that the value gets saved properly in your ticket.

    Tags

    Use tags to identify the purpose of this Handoff. Tags attach to the Zendesk ticket automatically.

    Separate each tag with a comma.

  4. Click Save.

Your Zendesk Handoff is ready to go.

Customer experience

The end user’s experience is seamless. Once they are handed over to the live support agent, the only thing that changes about their experience is who they are chatting with. AI Agent and live agent conversations remain in the same chat window.

End users can also minimize or close the chat window, regardless if it’s Ada’s or another brand’s chat environment, and expect their conversation history to persist; however, if they click the End Chat button, the conversation history clears.

Agent experience

When a Handoff to a live agent occurs through your AI Agent, the primary agent experience remains largely unchanged.

  • Agents still handle all Live Agent conversations from their Zendesk Agent Workspace dashboard.
  • Agents can also respond to end users asynchronously, and over an extended timeframe.

In all tickets except those created between June and August 2023, transcripts show AI Agent messages with the channel set to “messaging”:

Between June and August 2023, tickets originating from your AI Agent had the channel set to “Sunshine Conversations API”:

These settings update once an end user starts a new chat session. For AI Agents that have long persistence settings, you or your agents may wish to tell end users to clear their caches to display the updated channel source in the ticket.