AI category detection is a beta feature available to selected customers only. If you're interested in joining the beta, please email Pavel at pavel.kozlov@intercom.io
Overview
AI category detection enables Fin to automatically categorize conversations by topic, sentiment, or other chosen criteria.
Key benefits:
Automated categorization: Remove the need for customers to click through multiple options to specify their request before chatting with a support representative.
Improved efficiency: Reduce team response time and improve triaging accuracy.
Seamless integration: Use across all Intercom channels and in workflows, reporting, and inbox.
How it works
Use a conversation data attribute to create your categories and add an explanation, keywords, and sample questions to help Fin categorize conversations correctly.
Add the AI category detection step to a workflow and have Fin automatically detect categories based on the content of a conversation. These categories can be used in branches for seamless triaging and conversation routing.
Intercom’s patented AI Engine™ is responsible for selecting the most appropriate category based on the conversation’s content.
Step 1: Set up conversation data attribute for AI category detection
To enable AI category detection, you first need to create a conversation data attribute with a list of categories you want Fin to detect.
On the Conversations page, click Create attribute.
Click on the format dropdown and select List.
Fill in the "Name and "Description" fields for the attribute, e.g.
Topic Categories - Examply features categorized by topic.
Fill in the "List options" section with clear names and descriptions for each of the categories. You can click Add option to add more categories.
Click Save.
To use conversation data attributes for AI category detection, they must be in a list format and each option in the list must have both a name and a description.
Guidelines for effective AI category detection
Attribute lists used for AI category detection consist of two parts: a name and a description. The name should be a clear, short title, while the description provides detailed information which helps Fin identify this category in a conversation.
These categories can serve many purposes, from triaging conversations to detecting spam or analyzing customer sentiment.
Top tips:
Create clear, concise names - Choose short, descriptive names that immediately convey the category's purpose.
Write comprehensive descriptions - Include all relevant details about what belongs in the category. Think about every type of conversation that should fall under this category and describe them in the description. Providing a detailed description will help Fin categorize conversations correctly. You can include keywords and examples of customer questions.
Make categories distinct - Avoid creating categories that overlap too much. Your categories should be clearly different from each other, making it easy to determine which one best fits a given situation.
Try passing your category names and descriptions to a writing tool such as Claude AI or ChatGPT to define them clearly.
Real-world examples
Sentiment Categories:
Positive - A positive sentiment means the user who wrote the message seems generally happy or satisfied and is probably feeling a positive emotion.
Negative - A negative sentiment means the user who wrote the message seems generally unhappy or dissatisfied and is probably feeling a negative emotion.
Neutral - A neutral sentiment means the user who wrote the message seems to be neither happy nor unhappy and it is difficult to guess their emotion.
Spam Detection Categories:
Spam - Automated spam that is sent to the customer support agents. This topic includes auto-responders, newsletters, guest posts, and other general spam messages that could be ignored by the CS analyst.
Legit - Legitimate conversations in which the user has an actual issue that should be handled by a customer support analyst.
Topic Categories:
Projects - Projects are a collection of related tasks and activities aimed at achieving a specific objective or deliverable which can involve teammate collaboration, time tracking, milestones or goals, and status.
Billing - Billing encompasses managing subscription plans, invoices, payment methods, discounts, plan features, trials, account restrictions, refunds, and more for a seamless billing experience.
Account Management - Account Management covers discussions related to user accounts, including account creation, deletion, updating personal and payment information, and more.
Step 2: Add AI category detection to a workflow
After creating a conversation data attribute with your categories, you can start using it for AI category detection in workflows.
Go to AI & Automation > Workflows.
Create a new workflow or edit an existing one.
Add the AI category detection action anywhere in your workflow where you’d like Fin to categorize the conversation.
Select the attribute you created for your categories.
Now you can use the categories within your workflow. For example, you can add a branching condition based on an AI detected category and route the conversation to a specific team or teammate.
You can also add the AI detected category in a Note or a Message.
Save your workflow and then test the AI category detection using the workflow Preview.
Set your workflow live.
Teammate experience in the inbox
Teammates working in the inbox can see when a category has been detected by Fin in three places:
Conversation events within the conversation thread.
Internal notes.
The inbox sidebar (when conversation attributes are pinned).
If Fin sets the wrong category, teammates can correct it by selecting the appropriate category from the inbox sidebar. Please email us if this happens and we'll review it. 🧐
Optimize AI category detection
Start simple: Begin with a few key categories that cover the most common topics your team handles. You can expand and add more categories to an attribute as needed.
Refine your categories: Regularly review the categories being selected by Fin to ensure they're accurate and aligned with your customers' questions.
Try using categories in reporting: Experiment with using AI detected categories in custom reports to gain insights into the types of questions your customers are asking.
We welcome your feedback on how useful this! It will help us improve the feature moving forward.
Need more help? Get support from our Community Forum
Find answers and get help from Intercom Support and Community Experts