top of page
Writer's pictureNouran Smogluk

Zendesk Advanced AI: Best Practices to Improve Your Customer Support

With so many features, it’s easy to lose track of just how much Zendesk can do. Now, with the new Zendesk Advanced AI add-on, there’s even more potential to streamline your customer support and business operations.


For customer service leaders, Zendesk’s Advanced AI offers game-changing capabilities by using artificial intelligence (AI) to identify customer intents, handle incoming requests, and reduce the need for manual triage. With language models that can understand complex customer requests, this AI add-on enables your support team to respond faster and more accurately. to customer interactions. This cuts down resolution time and ensures a higher likelihood of delivering exceptional service at scale.


In this article, we’ll break down everything you need to know about Zendesk’s Advanced AI features to help you decide if it’s the right fit for your team. Whether you’re aiming to boost efficiency, improve response accuracy, or deliver a seamless customer experience, this AI-powered tool could be a key asset in reaching those goals.


What is the Zendesk Advanced AI add-on?


Zendesk Advanced AI is a set of AI features that are built for customer service teams. It’s offered as an add-on from the Zendesk Suite and Support Professional plans and is meant to help you enhance multiple parts of your customer experience–by optimizing your workflows, helping agents solve tickets faster, and automating certain parts of support interactions. 


What makes it a little hard to grasp is that Zendesk already some AI features in the same plans, called “Zendesk AI”:


  • AI-powered bots, which you can use to build your own chatbot. 

  • AI-powered guidance in the form of suggested macros so agents can select the relevant macro for a ticket faster

  • AI-powered insights in the form of content cues for help center article managers, which suggest help center content that might be missing based on exact keywords in tickets.


The Advanced AI add-on essentially offers a deeper and more complex set of features. We’ll go into all of them in more detail but here are some examples: 


  • Intelligent triage uses AI to automatically categorize and prioritize tickets. 

  • Generative AI for agents helps agents to write faster and more accurate responses to customers by suggesting relevant keywords and phrases.

  • Macro suggestions for admins helps admins to create more efficient macros by suggesting relevant actions and conditions.


Zendesk has a detailed breakdown of all features included in both.


Its potential advantages are obvious. You can use it to:


  • Improve agent productivity: Since it summarizes tickets and helps rewrite responses for agents, it should enhance their efficiency and save valuable time.

  • Reduce wait times. Both by improving agent productivity and by intelligently triaging and prioritizing more important cases, it should enable you to reduce wait times. 

  • Increase customer satisfaction. As with all AI features, when they really help you provide higher quality responses faster or offer immediate resolutions to problems, it can have a massive impact on customer satisfaction.


The key features of Zendesk Advanced AI


The add-on offers a number of features and the benefits of each one are slightly different. 


Zendesk Advanced AI is purchased as an add-on to your entire Zendesk instance and you will be charged per agent. The standard price is $50/month/agent. 


That means you get the most value out of it if all of these features are relevant to your needs or help you fill gaps in your current customer experience. 


Intelligent triage


Intelligent triage automatically predicts intent, language, and sentiment for tickets created via email on the web form. This is what it looks like in the Zendesk admin interface: 

Intelligent triage in Zendesk

Intelligent triage has a few important limitations:


  • It might not be a great feature for you if your primary support channel is chat. Because this intent prediction is a lot less reliable in shorter text snippets, it’s best suitable for teams that have heavy email volume.

  • Intent predictions have also been largely trained on industry data and Zendesk has a long list of industries you’d need to belong to, to get the most value out of it. That means it might not work well for you if big chunks of your support demand are caused by product-specific issues or require an understanding of terminology or jargon that’s only used in your business. 

💡Tip: If the out-of-the-box intents are not a fit, or you’d like to complement them, try Swifteq’s integration app between Zendesk and ChatGPT. You can use your own OpenAI account to detect any intent in tickets with minimal configuration effort.

That said, it can be a really powerful feature if you’re the right audience for it. That’s because:


  • It works across a huge number of languages. 

  • It allows you to measure sentiment on a scale of 5 instead of the normal 3 (from very positive to very negative). 

  • You can use its input in triggers to create very specific auto replies that try to solve issues independently. In theory, you could use it as an alternative to a fully-trained AI solution that interacts directly with your customers. 


Intelligence in the context panel


One of the ways intelligent triage is used is to provide additional context in a panel next to the ticket.

Context for Intelligence triage

The context panel briefly summarizes the most important information in the ticket and immediately suggests macros that might be relevant for that case. It also gives you a confidence rating so you can quickly estimate how trustworthy that recommendation is and adapt your response accordingly.


This is a valuable feature both for large teams that are handling high ticket volumes and small teams that have limited resources – it’s a simple and easy way to try and save your agents some time while they’re working on tickets. 


Autoreply and internal note trigger actions


Like the context panel, autoreply and internal note trigger actions are dependent on intelligent triage working really well.


You can create triggers that essentially say: If this intent is recognized, add an internal note or send this autoreply to the customer. 

autoreply triggers in zendesk

The use of the autoreply is obvious – to automate a certain number of tickets. It’ll likely be a bit more reliable than Answer Bot responses. 


There are a number of situations where internal notes could be very helpful for you.  


Say you have a large team with very different levels of knowledge and seniority. You might be trying to onboard a large number of people at the same time. Or you might just have a very complex product with tons of SOPs (standard operating procedures) that agents just have a hard time retaining. 


All of these could benefit from a note like the one above that reminds them of the most important things to watch out for. If intelligent triage just is not the right tool for your use case, there are always alternatives like our Zendesk ChagtGPT integration app


Macro suggestions for admins


Zendesk automatically analyzes your incoming tickets and suggests new macros to create based on the responses that your agents have been sending out.


macro suggestions in Zendesk

This is especially useful for a larger team, where many agents are likely to quickly create a personal macro (either in Zendesk or in their own notes) just because the process of creating “official” macros tends to take more time. 


It also means your team doesn’t have to consciously think about needing a macro for each case, because this happens automatically in the background. 


Generative AI for agents


This is one of those features that Zendesk decided to use an unhelpfully generic term for. 


Generative AI for agents is a feature that enables agents to quickly rewrite their responses based on predetermined prompts: 


  • Text expansion where an agent can write a short one or two-line response or a few bullet points and expand this. 

  • Change the tone of a response by making it more friendly or more formal. 

zendesk generative ai features

You can measure the impact of this type of feature on handling time, CSAT, or a quality score.


Generative AI for Help Center


Similar to the feature above, generative AI for your help center offers a similar set of features to rewrite text automatically and save time when it comes to creating articles. When paired with Zendesk’s other AI features like content cues, it can help you beef up your help center and have much more comprehensive coverage of the most important topics a lot faster. 


Best practices for implementing Zendesk Advanced AI


While it's still a relatively new Zendesk product, these are some top tips for ensuring you get the most out of Zendesk Advanced AI.


Establish clear goals for success


Since there’s no way to test it before investing the money and time to get it to work, the most important starting point is to understand exactly what you want to achieve with it. 


AI features should have a measurable impact on some of your metrics, and you should decide which metrics you want to target before choosing any AI solution. 


Brandon Tidd, Professional Services Consultant at 729 Solutions, said, “Previously the hype around Advanced AI was about bringing generative AI to the chatbot. Now the focus is on use cases where AI can help agents improve their efficiency.” 


The way you’d measure the success of an AI solution that interacts with your customers and one that’s primarily there to assist your agents is very different. 


The more precise your goals, the easier it will be to run a small-scale pilot program quickly so you can see if the potential is there.


Optimize data quality for accurate AI training


AI solutions rely heavily on data quality, especially when identifying customer intents and routing incoming requests. Before rolling out Zendesk Advanced AI, take time to clean up and standardize your historical data so that the AI has clear and relevant customer questions and common queries to learn from. Ensuring consistent data quality will help reduce errors, improve the accuracy of AI-driven responses, and ultimately contribute to a smoother implementation.


Map customer intents and complex tasks to specific workflows


Identifying common customer intents is a cornerstone of effective AI usage, especially with Zendesk Advanced AI’s language models. By mapping specific customer intents to clear workflows and actions, you can guide the AI to provide more accurate answers and reduce ticket resolution time. This approach ensures the AI’s responses align with real-world scenarios your team encounters.


Customize communication channels for AI interactions


If Zendesk Advanced AI is handling multiple messaging channels, such as chat, email, or social, make sure each channel is optimized for the AI’s capabilities. This might mean configuring specific language models or response styles tailored to each channel. For example, chat may benefit from faster service, while customer email requests might require more formal and detailed replies. Customizing these channels can help you meet customer expectations. 

 

Establish a feedback loop with agents


After implementing AI, make sure to enable ongoing training and gather regular feedback from your customer support agents to assess its effectiveness across a wide share of customer queries. These insights can help identify gaps in the AI-powered agents' capabilities. This will help you generate more relevant responses.


Plus, seamless collaboration with your agents increases buy-in as they see their contributions directly improve the system.


Invest in continuous monitoring and maintenance


A recurring pattern that many teams run into is to underestimate just how high the training and maintenance manual effort will be for any AI solution.


Jen Seran, Director of Operations at Stallion Express, said, “We undervalued the need for ongoing education for the AI system. We thought it would seamlessly integrate on its own but it needed consistent refreshment with fresh information to remain useful.” 


AI is not a plug-and-play system. It will take a very long time for it to run in the background without any major effort on your part – if it ever happens. Being prepared and having the resources to really invest in making it work is essential. 


Implementing tools that support your team


AI-based customer support solutions and tools like Advanced Zendesk AI can improve your customer service operations and shore up knowledge gaps, but it's crucial to ensure these tools genuinely support your human agents.


Setting clear goals, defining success metrics, and continuously refining your AI implementations can help you maximize the benefits of AI for both your team and your customers. 


But the journey doesn’t end there, and there are many ways to do that.


At Swifteq, we’ve developed a comprehensive set of apps—many of them AI-powered—that will help you streamline workflows, boost productivity, and create a better customer experience. It's the perfect complement to any AI initiative, including Zendesk's native Advanced AI.  



 




Written by Nouran Smogluk

Nouran is a passionate people manager who believes that work should be a place where people grow, develop, and thrive. She writes for Supported Content and also blogs about a variety of topics, including remote work, leadership, and creating great customer experience.


bottom of page