| Suite | Team, Growth, Professional, Enterprise, or Enterprise Plus |
Generative replies allow your AI agent to deliver automated, AI-generated answers in response to customer requests. These replies:
- Use information from your own trusted help center content.
- Give end users the information they need without having to leave the conversation to read an article.
- Don't require you to create or maintain custom answer flows.
Data shows that generative replies are typically three times more capable of returning relevant responses to customer questions as opposed to answer flows.
This article contains the following topics:
About generative replies in legacy AI agents for messaging
If you have an active help center, your AI agent can use generative replies to immediately respond to questions from your end users. These responses use generative AI to evaluate all the articles in your help center, then use that knowledge to provide concise, contextually relevant answers within the ongoing conversation.
When a relevant help center article is identified, generative replies are displayed as a single message. These messages are short (generally under 100 words) and reflect the AI agent’s persona. Any articles used to generate the response appear as links underneath the message.
Generative replies are always followed by a standard request for end-user feedback (“Was this helpful - yes/no”). You can configure how the AI agent responds to this feedback.
For generative replies, the language the AI agent uses depends on the language of the user’s initial message:
- If the user’s language is supported, the AI agent replies in that language.
- If the user’s language is not supported, the AI agent replies in the account’s default language.
For standard responses and messages sent as part of an answer flow (legacy), the AI agent uses the language set on the Settings tab.
If some or all of your help center is restricted, generative replies respect the view permissions set on articles. For details, see Using restricted help center content in AI agents for messaging.
If your account doesn't have an assigned intent model, your AI agent uses generative replies to respond to customer questions as soon as you activate generative replies. If your account does have an assigned intent model, your AI agent uses generative replies to respond to customer questions only in the following scenarios:
- An intent matching that question is configured to generate a reply.
- No intents are detected for that question.
For details, see Using generative replies with intents.
Activating generative replies
When you create a new AI agent, generative replies are activated by default. If you have any existing AI agents that don’t use generative replies yet, you can update the “If the AI agent finds relevant information” standard response so that the AI agent uses generative replies.
To activate generative replies in an existing AI agent
- In Admin Center, click
AI in the sidebar, then select AI agents > AI agents. - Click Manage AI agents for messaging, then click the AI agent you want to activate generative replies for.
- Select the Behavior tab.
- Click the If the AI agent finds relevant information section to expand it.
- Select Generate a reply.
- (Optional) Click Test AI agent, then enter phrases the AI agent can match to existing help center content to view samples of replies your end users might receive.
- Click Publish AI agent to apply your changes.
If your account doesn't have an assigned intent model, your AI agent now uses generative replies to respond to customer questions. If your account has an assigned intent model, proceed to the next section to make sure generative replies are active for all intents.
Using generative replies with intents
When you activate generative replies, replies are generated for questions that match intents without assigned answers and for questions without a matching intent.
Using generative replies for questions with specific intents
By default, any intent without an assigned answer will generate a reply. For each intent, you can leave this setting as-is (recommended), update it to not generate a reply, or add the intent to an answer in your AI agent for messaging.
To use generative replies for questions with specific intents
- In Admin Center, click
AI in the sidebar, then select AI agents > AI agents.
- Click Manage AI agents for messaging.
- Click the AI agent you want to manage generative replies for.
- Select the Intents tab.
- In the Answer column, click the drop-down to select an
option for each intent you want to update:
- Generate a reply (recommended) lets the AI agent create an AI-generated response to a customer’s matching free-text entry.
- Don’t generate a reply prevents the AI agent from responding to a customer’s matching free-text entry. Instead, it replies with the standard fallback response (“Sorry I couldn't answer your question”).
- Show an answer lets you connect the intent to an existing answer or create a new answer for it.
-
Click Publish AI agent to apply your changes.
Using generative replies for questions without matching intents
You can generate a reply for any question asked by an end user that does not have a matching intent.
To use generative replies for questions without matching intents
- In Admin Center, click
AI in the sidebar, then select AI agents > AI agents.
- Click Manage AI agents for messaging.
- Click the AI agent you want to manage generative replies for.
- Select the Intents tab.
- In the Questions without matching intents drop-down, select
one of the following options:
- Generate a reply (recommended) lets the AI agent create an AI-generated response to a customer’s matching free-text entry.
-
Don’t generate a reply prevents the AI agent
from responding to a customer’s matching free-text
entry. Instead, it replies with the standard
fallback response (“Sorry I couldn't answer your
question”).
- Click Publish AI agent to apply your changes.
99 comments
Jim Cochran
I'm seeing some bot interactions being transferred to agents immediately without the bot trying to generate a response to answer the question. We have articles that the bot should be able to reference. It's not happening all the time so the bot is configured corrected for generative replies. Any ideas on what could be going wrong?
0
Kevin Kilgarriff
Our articles are restricted to signed-in users. So my understanding is that if we configure Answerbot to have AI generate a reply, it won't source from anything that is restricted. However, I'm also seeing info that tells me it might be possible for that to happen if we set up JWT authentication. Is that accurate? If we set up authentication would any user who is authenticated have an AI generated experience that will pull info from restricted articles?
Additionally, am i correct that the authentication all happens behind the scenes? There's no extra step for the user, correct?
0
Lukas
Hi! Is it possible for the AI agent to learn from and provide links to articles outside of the Help Center? Specifically, can the 'Crawlers' section in the Help Center be used for this purpose?
1
Jakub Raduszewski
Hello!
If I have English (Singapore), English (United Kingdom) and English (Europe) language versions of my Help Center with different articles available for each of these regions, will bot only pick up the relevant region articles based on client's country?
“Every reply the bot generates is followed by a request for feedback (“Was this helpful? Yes/No”)” - can I see somehow in Explore what are the reply stats for this question?
0
PTS Agent
We previously installed the Zendesk widget on our Shopify backend using the widget code but did not create a bot at that time. Now, we are looking to switch to Zendesk's AI bot functionality. However, despite creating a new bot, I am still unable to find the "intents" tab.
Should I delete the previous widget configuration to fully enable the AI bot features? Additionally, the page you provided in the link (https://support.zendesk.com/hc/en-us/articles/4435802312730) no longer seems to be available.
I would appreciate your guidance on how to resolve this and properly activate the AI bot.
Thank you for your assistance.
0
Daniel Öhrström
Is it att possible to change the length of the generative reply? I feel that the responsive a little too short for it to provide enough information most of the times.
1
Emily B
Hi 1263082112129 regarding your response from last December 'Hi Havard, we're actively working on a solution to enable restricted content to be referenced for generative replies. We're expecting to release this in January 2024.' - is this something thats still being worked on?
Our company does not want the ‘read more’ button to be seen when using generative replies as we do not want customers to access the articles themselves, its for internal use only. However if we restrict articles, the generative replies do not work at all.
0
이지훈
hi Jordan Kolmer
If your team uses SSO to log in your customers, then it works.
Customers may or may not get answers from the AI agent based on their user segments.
But the weird thing is, when I test this with my test account that doesn't use SSO, it doesn't work correctly and the AI agent answers all the questions as if it doesn't know. Even though my test account has the user segments set up correctly.
I concluded that only the end users who use SSO and have user tags set up through SSO and thus have the user segments set up, are getting the correct answers from the AI agent.
0
Jordan Kolmer
I require users to sign into my help center before they can see any content. The generative AI replies do not ever return a result. Was the functionality released to include content from restricted articles based on the signed in user's segment?
0
Sally Anne Dishong
Thanks for your response, Thai! Our Customer Support team is currently evaluating the use of the bots and your experience is helpful as we continue our research.
0
Sign in to leave a comment.