This article provides a quick reference to the Editions, Data Centers, Languages supported by Zia, and a short overview of all Zia features available in Zoho Desk.

Intelligence  tools

Zia insights  

Read more about Zia Insights

Sentiment analysis

When agents handle a high volume of tickets each day, a generic reply like “I understand how frustrating this can be”, "Our customer support executive will reach out to you", or "We are sorry for the inconvenience caused" can fall short. It can come across as impersonal and fails to truly understand what the customer is experiencing and offer them empathy. 

But what if agents could better understand the customer’s mood and craft responses accordingly? That’s where Zia can help.

Zia analyzes the content of each incoming message and categorizes the sentiment as positive, negative, or neutral. It also highlights the key phrases or keywords that influence classification . This allows agents to identify and prioritize tickets with a negative tone and respond in a way that is empathetic and with better awareness of the context.

Note: Sentiments under Zia insights is powered by Generative AI whereas, sentiment analaysis is powered by Desk algoritham which is used in automation process and tracked under Zia Dashboard. 

 Read more about Sentiment analysis 

Auto tag

Tags help organize and categorize tickets based on certain keywords used in the tickets that are distinct to the ticket, for example verification, credit card, request, documentation, etc.With Zia, this process is automated as Zia analyzes existing tickets to detect common keywords, groups them into clusters, and assigns relevant tags to each cluster.

When new tickets arrive, Zia matches keywords to these clusters and auto-tags the tickets accordingly. It continues to add tags as the conversation evolves, without removing the previous tags, allowing multiple accurate tags to be added to each  ticket. 

Agents can also view the auto- tags that are trending for the past 24 hours in the predictions dashboard, under Zia dashboards in the Analytics module. 

This allows agents to spot trends from the dashboard, identify common issues, and prioritize responses. For example, if an agent spots a spike in tags like "refund" or "return" under trending auto tags, this could indicate service concerns and help teams respond proactively.  

Read more about Auto Tag.

Thread level keywords

Zia automatically generates keywords for each customer response (thread) within a ticket, helping agents quickly understand the core context of the entire conversation. This is especially useful for tickets that involve multiple teams or lengthy back-and-forth discussions, as well as in cases where the ticket has been shared with other agents who can find it challenging to draw context from long conversations.

Using thread-level keywords agents can:

1. Track key developments that happened in the conversations
2. Automate alerts, field updates notify specific teams by setting the workflow rule criteria as keyword contains
3. Achieve automatic ticket routing
   

Read more about Thread level keywords.

Prediction tools

Zia notifications for Anomalies and Alerts

Zia proactively monitors your help desk for anomalies in ticket traffic—like sudden surges or dips—as well as alerts related to system activities such as Answer Bot training, Auto-tag creation, and Annotated ticket status updates. When triggered, a red icon appears in the notification panel at the bottom of the screen to bring these insights to your attention.

1. Anomalies – Unusual changes in incoming or outgoing ticket responses.
   
2. Alerts – Updates on model training, auto-tag creation, and usage of annotated ticket statuses. 
   

Note: Notifications remain available for 15 days from the time they are notified by Zia.

Anomaly prediction

1. Quickly inform the product and QA teams regarding the issue.
2. Send proactive updates to affected customers to ensure smooth communication.

1. Navigate to Setup > Zia > Anomaly.
2. On the Zia Anomaly page, select the department from the drop-down and toggle to enable the feature. 
3. Under the Predictions for Incoming Tickets tab/ Predictions for Outgoing tickets:
1.  Select the minimum number of incoming/ outgoing tickets for Zia to consider as an anomaly 5,25,50. OR
   
2. Customize the trigger criteria to specify the permissible deviation percentage for different ticket response volumes. [For example, imagine that you set 50 as the deviation percentage for incoming ticket responses, and the trend line predicts 100 responses at a given hour. In this case, Zia will only trigger an anomaly when the actual incoming ticket responses are 150 or higher at that hour.]
4. Choose when to receive the notification by enabling- Notification Settings for Business. 

Field prediction 

In Field Prediction, Zia trains and learns from past ticket data to identify patterns and automatically populate picklist fields like category, priority, or issue type. This helps agents by reducing their manual efforts of entering field values manually as Zia auto-fills the predicted values in the fields. Which is used for auto triaging tickets to the right department, right agent, or team reducing the resolution time.

Read more on field prediction.

Summarization tools  

Ticket summary

Zia provides a quick summary of the latest conversations  in a ticket, condensing key points into bullet form for easy understanding. Agents can filter by conversation type (incoming, outgoing, forwarded, public comments, private comments), regenerate summaries for clarity, or translate them into supported languages.

Ticket summary takes into consideration the latest 30 conversation with Zia GenerativeAI model enabled ,and the five most recent conversations when using ChatGPT. 

Read more on Intelligent Ticket summarization.

Thread summary

When a customer shares multiple concerns in one message, the Summarize Threads feature extracts and highlights the main points. This helps agents grasp complex queries faster and respond more effectively.

Read more on thread summary.

Generative AI capabilities

GenAI LLMs 

Zoho Desk offers two Generative AI models:

1. Zia: Zia's generative AI abilities are powered by open source language model (Llama 3.1 and Qwen 2.5). Zia generates responses exclusively based on the knowledge base (KB) articles for Answer bot and Reply assistance. This approach offers a more secure solution, making it ideal for businesses that prioritize data privacy or require responses to be limited strictly to verified, internal content.
2. ChatGPT: ChatGPT uses openAI capability to analyze the ticket content and generate response accordingly. It makes use of the information available in the knowledge base as well as the open domain, based on the configuration, to generate responses. 

Answer Bot 

Answer Bot is the AI-assistant that goes through the information repository to find the best fit information that assists you in solving the query in hand. Agents and customers both crave a direct answer to the question in hand, rather than going through an entire article to fetch the information. This is where Answer Bot comes into play. 

For example, if a customer wants to know the required documents for a visa appointment, instead of reading through the entire help article, they can simply ask Answer Bot. The bot will scan the article and list the required documents directly.

A customer raises a ticket asking how to troubleshoot an issue in the latest version of the software. Instead of manually searching through multiple help articles, the agent can simply ask Answer Bot the specific question. The bot scans the help documentation, retrieves the most relevant solution, and provides a direct answer along with the article link, which the agent can use.

While agents can access the bot from the ticket detail view, customers can leverage the bot through the help center, websites, and landing pages where the bot is deployed. 

Read more on Answer bot.

Reply assistance 

Reply assistance helps agents draft responses by analyzing customer questions and pulling relevant information from Knowledge Base articles and open domain. If no relevant KB articles are available, Zia prompts the agent to add information and retrain the system.

For example, if a customer reports a technical error, Zia suggests a troubleshooting response and links the corresponding article for quick reference.

Agents can:

1. Regenerate responses to adjust tone, length, and language
2. Edit or copy responses directly
3. Simplify technical jargon into easy-to-understand language
   

Read more on Reply assistance.

Generate content

Agents can generate content on demand using simple prompts to draft custom replies, emails, and invitations via simple prompts. For example, an agent can prompt:

"Generate an invitation email for the upcoming end of season sale starting from 20th of December to 2nd of January."

Zia uses Knowledge Base and open domain data to draft the content. Agents can edit, regenerate for different tone, length, and even translate it into Zia supported language before using it in reply. Read more on Generate content.

Writing assistance

Agents can fine-tune their drafted responses with Writing assistance by selecting specific sentences to modify. They can adjust the tone, length, and language to suit the context better. For example, agents can make use of writing assistance to match the tone  or language of the customer reply. Read more on Writing assistance

Content analysis

Content analysis helps agents ensure replies are clear and customer-friendly. It offers grammar suggestions, spelling checks, readability scores, and suggestions to simplify complex sentences. For example, in a long response explaining multiple steps, it can suggest splitting content into shorter sentences for easier reading. Read more on Content analysis