On August 1, we hosted an engaging event on "How to Prepare your Team for AI Driven CX." Attendees learned best practices for involving teams in AI adoption and heard valuable insights from Heather Eeles, VP of Client Support at Benevity, who shared her experiences with implementing AI tools. The session highlighted the importance of engaging team members to ensure smooth integration and maximize the benefits of AI. Below, you'll find resources, the event recording, and answers to the top questions from our Q&A session. Don't miss out on these valuable takeaways to help prepare your team for an AI-driven future.
Event Recording
Q&A Summary
Q1. Is there a way to train the AI to speak in our company's language? We use unique terms for our customers and also have words we do not want used when replying to customers.
Yes, you can use bot personas to determine the style of expression applied to AI-generated messages, giving them a consistent voice that reflects your brand.
Resources: Using bot personas to add personality to AI-generated responses.
Q2. What should we prepare our agents for when it comes to utilizing AI?
To prepare for Zendesk AI, it's important to educate your agents on its capabilities through training and test environments.
Q3. Our teams add tables to our comments, such as a 4x7 table with products and costs. Will these tone change features alter our table data?
It is possible that the generative AI options may alter your data table in a ticket reply, so we recommend that you highlight the text you want improved by generative AI so the table is not considered at all.
Q4. Our teams add tables to our comments, such as a 4x7 table with products and costs. Will these tone change features alter our table data?
It is possible that the generative AI options may alter your data table in a ticket reply, so we recommend that you highlight the text you want improved by generative AI so the table is not considered at all.
Q5. Can I have a bot just visible to Agents for asking AI for answers such as in ChatGPT?
If you have an internal Help Center, or one that is only available to agents, then certainly! The AI Agent generative replies will only pull from content within your Help Center at this time, and does not index the same breadth of information that ChatGPT does. There is also an EAP for Agents to index this information from within Agent Workspace.
Resource: Generative search for Agent Workspace (EAP)
Q6. How do you use sentiment to change ticket routing?
You can route triaged tickets using omnichannel routing. Further, intelligent triage automatically determines the ticket intent, the language it's written in, and the customer sentiment (positive or negative). You can then take this information and route tickets from email, messaging, and calls.
Resource: Routing automatically triaged tickets using omnichannel routing
Q7. With generative replies for the bot, we have been hesitant to implement them because we can't test them in our sandbox. I'd like some ideas of what will happen. If the bot generates a reply for the customer, will it still put them into a flow, or what does that look like?
Within Bot Builder, you can designate when you want a generative reply to occur. If the bot detects a matching answer within your Bot Builder, then that will always take precedence.
Resource: Using AI to generate replies in a conversation bot
Q8. For generative responses, it sounds like the responses will only be as good as your Help Center, so keeping it up to date is critical. Is there a criterion or failsafe that the bot uses to identify bad information that might have slipped through the cracks? For instance, information from an article that hasn't been updated for a certain time.
Yes, responses will only be as good as the quality of the Help Center content. While the bot does not have any inherent failsafe components, other tools within Guide, such as Article Verification and Content Cues, can help content administrators keep article information up to date and valid for end users.
0 comments