Powered by Zendesk AI, AI agents are designed to fully resolve customer issues, no matter the complexity. As AI becomes increasingly essential to customer interactions, AI agents enable you to move toward greater automation and efficiency.
This article provides an overview of the admin tasks to set up and start using AI agents functionality for conversational bots and autoreplies in email and web form support requests.
Related articles:
- Overview of AI agents
- About automated resolutions for AI agents
- On-demand: Learn how to use AI agents (Free training course on AI-powered intents and autoreplies)
Prerequisite: Configure a basic messaging channel
If you plan to create a conversation bot, before you can do so, you’ll need to give it one or more channels where it can communicate with your customers. Messaging is the most commonly-used. You can offer messaging through a Web Widget for your website or help center or through mobile SDK for your Android or iOS app.
If you haven't already configured a basic messaging channel, see Getting started with messaging.
Optimize your help center content for AI agents
AI agents can recommend articles from your help center and use generative AI to create replies based on your help center content. You'll want to make sure your content is optimized to return the best articles and answers.
If you're just getting started with your help center, you can review tickets and other resources to find issues to populate your knowledge base. Whether you are writing new content or updating exiting content, there are some best practices you should keep in mind to optimize your content.
To optimize your help center content for AI agents, see Optimizing your help center content for bots and autoreplies.
Create an AI-powered conversational bot
You can quickly set up a conversation bot without a lot of customization. You'll need to set up a channel for the bot, create the bot, then publish the bot to the channel to make it live for your customers.
Set up an AI-powered conversation bot
You can set up a conversation bot that responds to your customers using your existing knowledge base. Then take the time to design more custom responses.
When you create your conversation bot, you can:
- Choose a name that makes it clear customers are not talking to a human.
- Add a custom avatar to reinforce your brand or use the default icon.
- Set your language. By default, the language is the same as the default language for your account. Later, you can configure the bot for multiple languages, if needed.
- Use generative replies for a quick start. With a help center and generative replies, your bot can respond to questions in a conversational manner, providing responses pulled from relevant knowledge base articles.
- Give your bot a personality. Select a persona to reflect your organization’s tone and brand.
To set up a conversation bot, see Creating a conversation bot for web and mobile channels.
Make the bot live in your messaging channel
At this point you have a fully functional bot you can make live and use without any other configuration. You need to publish the bot to make it live in your messaging channel. You can do this now or you can hold off and do it later.
If you make the bot live now, you can monitor requests and find common issues to customize your bot and make it more effective. If you already have some ideas in mind, you can hold off on publishing and customize your bot now.
To publish your bot, see Adding a bot to a messaging channel.
Customize your conversation bot with answers
After you’ve created your conversation bot, you can customize the bot with answers, if you want. With answers, you can create customized conversation flows to offer the best responses and guide customers through conversations that require more direction. After you create answers, you'll need to publish (or republish) the conversation bot to start using the answers.
Create answers for questions customers might ask
An answer addresses a single question your customers might ask about. A conversation bot can have as many as 500 answers.
There are two ways you can create an answer:
- Use a template as a starting point
- Create an answer from scratch.
An answer consists of a sequence of steps that are built into a template as out-of-the-box behavior or that you create and configure from scratch. Each step represents a message the bot can show to your customer during a conversation.
- Include single or multiple steps in your answers. Examples of steps you might include are branching based on conditions, transfer to agent, or make an API call to an external system.
- Enhance answers with rich content. Add emojis, links, images, and gifs to create a more engaging experience for customers.
- Add training phrases to help the bot understand customer questions.
- Assign intents to answers to help your conversation bot match and deliver the most relevant answer for a question. If you have an intent model assigned, you can use intents instead of training phrases. Intents are predetermined topics that represent customer questions and requests. When you assign an intent to an answer, the bot knows to show this answer whenever it identifies the intent in a conversation with a customer.
Before you start creating answers, be sure to review the complete list of step types that are available and be familiar with the best practices.
To customize your conversation bot with answers, see Creating conversation bot answers for common questions.
Publish your bot to start using your answers
Before your answers are available to your customers, you must publish the bot. Answers are not published individually. When you publish the bot, all answers that are in the ready to be published state are also published.
To publish your bot, see Adding a bot to a messaging channel.
Extend AI agents to email and web form
You can add autoreplies to customer notification responses to help customers self-solve their requests. Autoreplies with articles use triggers to determine when the reply is sent and what information is included in the email or web form response, and add up to three links to relevant knowledge base articles.
To add AI agents to your email and web form:
- Use email autoreplies to create triggers that include help center articles in email responses to help the customer resolve their issue.
- Use autoreplies for web forms to do the same in web form responses through your help center.
Monitor AI agent performance to improve efficiency
After you've set up AI agents functionality, you can monitor performance and continue to improve efficiency using pre-built dashboards and the Insights dashboard.
- Analyze autoreply performance in email and web form notifications. Identify opportunities to improve suggestions, resolutions, and click-through rate.
- Monitor bot builder activity for more granular information about bot and answer performance.
You can also monitor the Insights dashboard, as soon as 48 hours after launching your bot, to help you identify customer questions that aren’t getting useful responses.
- Use the Insights dashboard for a comprehensive overview of metrics, response performance, and coverage gaps. It can help you optimize your bot configuration and improve the self-service experience for your customers.
- Review conversation transcripts. You can use conversations that resulted in automated resolutions to understand effective answers. You can also review unresolved conversation transcripts for patterns and similarities in less-successful interactions.
- Find commonly unanswered questions with intent suggestions. If you have an intent model assigned, you can improve existing answers by adding suggestions or creating new answers to incorporate into a conversation flow.