On September 26, the Zendesk Community hosted an event that explored how to enhance customer service by transitioning from basic Chat to Zendesk’s AI-powered Messaging. Participants learned about methods for upgrading simple chat interactions to more advanced, automated communication techniques. The session provided practical insights into using AI and machine learning to improve customer support, including the benefits of offering 24/7 assistance and personalizing customer interactions.
The event also covered the integration of Zendesk’s messaging tool across various channels, such as web, mobile, and social media, to ensure a consistent service experience. Attendees discussed the role of real-time analytics in gaining insights into customer interactions and enhancing overall support performance.
For those interested, we encourage you to access our event recording, Q&A highlights, and presentation materials. These resources are designed to support anyone looking to improve their customer service capabilities with AI-powered messaging.
Event Recording
Q&A Summary
Yes! Enabling messaging features helps us direct you to the right resources quickly, allowing you to get answers faster. We also prioritize urgent chats, ensuring that high-priority inquiries receive immediate attention. This leads to a smoother and more satisfying experience overall.
Yes, as of April this year you have the ability to request intents in the admin center. The evaluation from Zendesk will take up to 2 weeks.
Resources: Announcing the Custom intents: Request new intent beta
Yes, messaging is included in our Suite plan, just like Chat. However, while AI can assist with resolutions, there may be costs involved depending on the specific AI agents used, like ADA or Zendesk. For more details on pricing, we recommend reaching out to your account manager.
Resources: Getting started with Zendesk AI and Advanced AI, About automated resolutions for AI agents
To successfully use both Chat and messaging, start by identifying the right channels for messaging, such as social media and native integrations. Begin by creating separate messaging brands and gradually roll them out. This allows you to see how messaging performs in real-time and refine your workflows to better suit your needs.
Resources: Welcome to messaging
Zendesk can detect a user's preferred language from their Web browser preference setting. The accept-language header, which is passed via HTTP, contains information about the user's language preference. If that is present, the language can be detected.
Resources: Adding multiple languages to Zendesk Support, Understanding and activating live conversation translation
Proactive messaging can be triggered by user actions on the page, like clicking a URL or button. For more advanced interactions, we have APIs that enable these features easily.
Resources: About proactive messages
Yes, there are AI intents for proactive messaging. The effectiveness largely depends on how you set up your bot. For example, you can prompt users by asking if they have questions if they linger on a page too long. By guiding them with structured messages, you can encourage engagement and help them find the information they need. This approach can significantly boost user interaction with your bot!
Resources: Creating conversation bot answers for common customer questions
Yes, you can link custom fields with AI agents. In the board builder, there’s an option to ask for details where you can map up to eight custom ticket fields. These fields are filled out by the user during the conversation and will be populated when the conversation is handed over to an agent. It works just like standard custom ticket fields.
Resources: Understanding the step types for bot answers
Our ML work is guided by the following principles:
- Training data must be de-identified to remove information such as email addresses and numbers from our models and annotation resources
- Customers must be able to opt out of training
- We respect data deletion
- We respect data locality in accordance with our Regional Data Hosting policy
Resources: Zendesk Trust Center
To implement AI-powered messaging effectively, focus on two key areas:
- Bot Configuration: Start by setting up your bot to handle most support tasks. This includes enabling auto-replies and utilizing AI features to ensure your bot is available 24/7.
- Agent Productivity: Enhance your agents' efficiency with advanced AI capabilities, such as improved options in the messaging workflow.
By combining these strategies, you can streamline support operations and optimize your workflows for better performance. Starting with the bot is a great way to kick things off!
Resources: Getting started with AI agents using bots and autoreplies, Overview of AI agents, Getting started with Zendesk AI and Advanced AI
Yes, you can release messaging to a specific segment of customers. However, it's recommend considering a broader approach by including all customers. You can customize the experience based on their status—like offering paying customers the option to chat with an agent while providing trial users with helpful articles. This way, you ensure that all customer segments feel valued and supported.
The AI uses knowledge base articles to answer customer questions by focusing on intent matching rather than just keyword matching. This means it understands the context of questions, even if they're phrased differently or contain misspellings, making it effective in providing relevant responses.
Yes, you can view the conversations customers had with the chatbot even if they don't escalate to an agent. Simply go to the bot's configuration in the insights dashboard to access transcripts of automated interactions. Additionally, our QA system can review both agent and bot interactions, providing another way to monitor those conversations.
Resources: Review bot conversation transcripts
Yes, bots retain the context from previous conversations when answering new questions. This means that even if the conversation is happening within a specific flow, the bot can remember earlier interactions, ensuring more relevant and coherent responses.
AI Agents is a large part of AI-powered messaging. It has that bot functionality to deflect tickets for common issues and hand over tickets to agents when they are more complex. But AI agents isn't the only part of AI and automation within messaging. Messaging also has automation with integrations within Agent Workspace. More specifically, omnichannel routing as an example.
Yes, our roadmap for AI agents is constantly evolving to stay competitive. One exciting feature coming soon is quality assurance (QA) for AI agents, which will help analyze their performance. We have several new features planned for the next few quarters, so I recommend speaking with your representative for more details. Additionally, you can apply to attend our AI summit on October 9th, where we'll discuss our upcoming AI features and advancements!
Currently, to use the AI agent features that access knowledge base content, you do need to have a Zendesk help center. However, you can link to an external knowledge base in other ways, such as a button in a bot answer.
Using the AI summary as a placeholder is not possible at this time, but we do have a relevant feature request.
Resources: Add AI Summarization as a placeholder to embed in Macros and Comments
Information on data usage can be found here: Zendesk AI Data Use Information
The second brand for Messaging won't automatically pull from your main guide or knowledge base. However, you can set up Messaging similarly to Live Chat, allowing you to effectively use both for different teams. This way, you can enjoy the benefits of both channels!
The best way to make the bot smarter is to use training phrases. We recommend using as many multiple word training phrases as possible per answer, avoiding single word phrases.
Resources: How can I improve my messaging bot to find the match I expect?
Information on migrating help center content can be found here: Migrating existing content to your help center.
Custom CSAT was just released!
Resources: Announcing the customizable CSAT experience
Yes!
Resources: Configuring messaging responses and business hours
Event Resources
0 comments