AI and Generative AI are the driving forces currently reshaping the future of CX. While exciting, it also brings disruption and a ton of new pressure for CX teams and leaders. At our last community event, we covered our newest Generative AI advancements and how Zendesk continues to infuse them into all parts of our product portfolio. We also covered best practices and hosted a live Q&A with our product experts.
Video Recording
Top Questions from the Q&A
Q1: How do we measure success for AI? Is there a possibility of generating a report that differentiates the impact of AI on various KPI metrics, pre and post-AI implementation?
There are a few approaches here, but first, decide which KPIs matter to you (first reply time, full response, etc.). Then slice those metrics against the intent/sentiment/language you have used to power your intelligent workflows - or you could even have your intelligent workflows add a tag or other indicator to allow you to compare AI-assisted vs non-AI-assisted.
Q2: Is it necessary to use the knowledge base to manage correct answers to the bot?
For the Generative AI chatbot, the answer today is yes, but it is in EAP, so this could evolve. Alternatively, you can also configure pre-trained answers where you would control the response more with bots.
Resources
Using AI-powered intents with conversation bots
Q3: Where does Knowledge Base Generative AI pull its language core for creating articles? Many of our articles are product-specific.
We're leveraging OpenAI. In this scenario, you would want to include the product-specific names in your outline or initial content you write. Or you can modify the content after it is generated to include the specifics that are just right for you. We've been using the tool to take an outline with a few bullets to a very solid rough draft that can then be easily enhanced from there.
Q4: What is the AI feature that tags tickets with the intent?
Intelligent triage! These tools are all useful on their own or in combination.
Resources
About intelligent triage and Intelligence in the context panel
Q5: Is Generative AI included with Zendesk Professional Suite?
Generative AI features are not tied to plans, but you can see the detailed requirements in our summary article linked below.
Resources
Zendesk Generative AI EAP capabilities overview
Q6: If I already have flows set up, will this conversational AI void my existing flows? Will I need to revise and/or delete the existing flows that I manually created in order to maximize the effectiveness of generative AI?
The bot will always first try to find an answer flow to first match a customer's intent. If it doesn't find one that matches, it tries to answer with generative replies. So you will not need to delete any of your existing flows. And now, with the intent dashboard, you can control what intents are answered using generative replies or a certain bot flow
Q7: How does the Zendesk AI handle more complex requests? For example, more troubleshooting for issues and not simple FAQs. Can the bot ask the user to clarify or start with one topic and then dive in further?
Yes! We have disambiguation prompts in place where the bot is unable to understand clearly what the customer is asking for. This is an Advanced AI capability that gives the bot the capability to confirm what the customer's intent is so that it can offer the exact help the customer needs help with
Q8: If AI is the one interacting during chat/messaging, is there a way for our Agents/team leads to monitor a live interaction?
We want to allow customers to self-serve to save agents time, which is why agents cannot editor monitor bot responses generated by GenAI.
Q9: Does the smart conversation also work for other languages?
Yes! Generative replies work in 20+ bot languages.
Resources
Languages supported in Zendesk bots
Q10: How long does it take for Macro Suggestions to start suggesting macros? We have enough volume and have been intentionally not using macros on similar tickets to urge AI to prompt a macro creation, but have not seen one yet.
Macro suggestions for admins are refreshed at the beginning of each month. But if you referring to suggested macros for the agent feature, we update the models about once a month. Once you're eligible, you may have to check your account settings to enable the feature when you are eligible.
Resources
Creating macros from macro suggestions for admins
Enabling and disabling suggested macros
Presentation Material:
Future Events & Opportunities