Add-on | Copilot |
The entity detection feature, which is part of intelligent triage, allows you to define and detect unique information in tickets and messaging conversations. You can then use that information to populate associated custom ticket fields and power automated workflows and reports.
Entity detection is similar to other intelligent triage predictions in that it allows you to automate workflows based on information automatically detected within customer requests.
This article contains the following topics:
- About entity detection
- Creating and configuring a custom field to represent an entity
- Configuring ticket field update behavior and entity highlighting
- Turning off or deleting a custom field associated with an entity
- Taking action on a highlighted entity in a ticket or messaging conversation
Related articles:
About entity detection
Customer requests often contain specific pieces of information that your agents need to identify and act on in order to solve the request. Entity detection makes this process faster and easier by automatically identifying this critical information and making it available in automated workflows and reporting.
An entity is a piece of information that is meaningful for your business, such as a product name. To detect an entity, you:
- Create a custom field to represent the piece of information (entity) you care about. For example, you might create a custom field called Product Line.
- Associate values with that custom field to represent details about the entity. For example, you might associate the Product Line custom field with the values Camera Model A, Camera Model B, and so on.
The custom field and its associated values can then be used in routing, automations, and reports to optimize your workflows.
Additionally, entity values are highlighted in blue in tickets, making it easier for agents to scan the ticket for important information.
Technical information about entity detection
When you create a custom field to represent an entity, the system adds that custom field to any tickets where a matching value has been detected. You must add the custom field that represents the entity to your ticket forms if you want agents to be able to view the field.
When tickets are created or updated, the custom field may be populated based on the entity value detected in the ticket, depending on how you've configured the ticket field behavior.
When the field is populated, a tag is also automatically added to the ticket to help you build triggers, automations, and reporting. These tags reflect the entity values you configure and are structured as entity_<entity custom field name>_<entity value>.
Also, note that an entity’s value can change over the course of a ticket’s lifecycle. For example, a ticket that starts out about one topic might evolve into another. Because of this, you can configure whether and how the custom field is automatically updated throughout the ticket’s lifecycle.
When configured, entity detection can detect entity values even when there are misspellings, including switched or missing letters. Detection of misspelled values:
- Works only when the first letter is the same as the entity value.
- Doesn’t work on patterns shorter than 6 letters.
- Allows for up to two errors in a word, including added, missing, misplaced, or replaced letters.
- Might not work in languages without spaces, or where a single character change results in a completely different meaning.
Entity detection works on all channels supported by intelligent triage. It’s not possible to turn entity detection on or off for only a subset of channels. Additionally, entity detection works on ticket subjects, public comments, and internal notes. Highlighting works only on public comments and internal notes.
Understanding AI-powered prediction
Entity detection is part of intelligent triage, an AI-powered feature that makes predictions to help you automate your workflows. However, entity detection accomplishes this goal in slightly different ways than other intelligent triage predictions. Intelligent triage predictions can be broken down into two types:
- Classifiers: Prebuilt by Zendesk and used in intelligent triage. Classify tickets into specific intent, language, and sentiment taxonomies based on our machine learning models.
- Extractors: Created by you and used in entity detection. Detect and extract specific information based on your configuration.
You can see and manage all predictions in Admin Center.
Creating and configuring a custom field to represent an entity
Admins can create a custom field to represent an entity in Admin Center. When you create a custom field, you must configure its corresponding values. For example, if you create a custom field to represent an entity called Product line, the values might be Camera Model A, Camera Model B, and so on.
The steps will differ depending on which type of field you create.
To create and configure a custom field to represent an entity (drop-down and multi-select)
- In Admin Center, click
AI in the sidebar, then select Intelligent triage > Entity.
- Click Add entity.
- In the Name field, enter a descriptive name and click Next.
- Under Select field type, select either Drop-down or
Multi-select.Tip: To create a Regex field, see the steps below instead.
- Click Create.
- On the Values settings tab for the entity custom field, click Add
new to manually add individual taxonomy values.
You can also click Upload CSV to upload a CSV file of all values.
- To add values manually:
- Click Add new.
- Enter the value that should be detected in tickets.
- Click the checkmark.
- To upload a CSV file:
- Click Upload CSV.
- Click Choose a file or drag and drop here to select a CSV file, or drag and drop the CSV file into the field.
- Click Upload.
- To add values manually:
- Click Save.
Next, you'll configure ticket field update behavior and entity highlighting.
- In Admin Center, click
AI in the sidebar, then select Intelligent triage > Entity.
- Click Add prediction.
- In the Name field, enter a descriptive name and click Next.
- Select Regex.Tip: To create a Drop-down or Multi-select field, see the steps above instead.
- Click Create.
- On the Entity matching rule page, enter the Regular expression
(regex).
These are the regular expressions that your entity uses to identify the relevant content in tickets. For example, "([1-9]|0[1-9]|[12][0-9]|3[01])\D([1-9]|0[1-9]|1[012])\D(19[0-9][0-9]|20[0-9][0-9])", which is a common regular expression for detecting dates.
Note: Expressions should not start with^
or end with$
delimiters. The presence of these delimiters is probably a sign that this regex was created for validating inputs, not for matching sub strings in longer text. Instead, consider using/b
as a delimiter in case you need it.You can enter text in the Test detection field in the right panel to check if the entity will be detected.
- Under Update ticket field with detected values, configure the
behavior of the ticket field associated with the entity custom field by
selecting one of the following options:
- Don’t update ticket fields: The ticket field is not automatically populated with any value. An agent must manually populate it by clicking Update.
- Values in first message only: The ticket field is populated with the value detected in the ticket’s first comment or conversation’s first message. Subsequent comments or messages do not automatically update the ticket field value.
- Values in subsequent messages only: The ticket field is initially populated and subsequently updated with the value detected in any ticket comment or message except for the first one.
- Values in all messages: The ticket field is initially populated and subsequently updated with the value detected in any ticket comment or message.
- (Optional) Under Agent tools, select Highlight entity values in
all messages if you want entities to be highlighted in blue in
tickets and messaging conversations.
When this option is selected, agents will also be able to take action on highlighted entities.
- Click Save.
Configuring ticket field update behavior and entity highlighting
You can configure how the custom ticket field associated with the entity is updated based on ticket comments or messages. You can also configure whether entity values are highlighted in tickets and messaging conversations.
To configure ticket field update behavior and entity highlighting (drop-down and multi-select)
If you just configured the values for a drop-down or multi-select entity, skip to step 3.
- In Admin Center, click
AI in the sidebar, then select Intelligent triage > Entity.
- Click the entity custom field
you created above.
The General settings tab opens.
- Under Update ticket field with detected values, configure the
behavior of the ticket field associated with the entity custom field by
selecting one of the following options:
- Don’t update ticket fields: The ticket field is not automatically populated with any value. An agent must manually populate it by clicking Update.
- Values in first message only: The ticket field is populated with the value detected in the ticket’s first comment or conversation’s first message. Subsequent comments or messages do not automatically update the ticket field value.
- Values in subsequent messages only: The ticket field is initially populated and subsequently updated with the value detected in any ticket comment or message except for the first one.
- Values in all messages: The ticket field is initially populated and subsequently updated with the value detected in any ticket comment or message.
- (Drop-down only) If you selected any option except Don’t update ticket fields, select Replace existing ticket field value if the new value in the ticket field should overwrite the old value. Otherwise, the ticket field will not be updated if it already has a value.
- Under Agent tools, select Highlight entity values in all messages if you want entities to be highlighted in blue in tickets and messaging conversations. When this option is selected, agents will also be able to take action on highlighted entities.
- Under Detection settings, select Detect misspelled values if you want to allow the system to find misspelled entity values.
- Click Save.
Turning off or deleting a custom field associated with an entity
If you no longer want an entity to be detected in tickets, you can delete the custom field associated with the entity. When you do this, entity-related highlighting is removed from all tickets, even tickets that were created while the entity was active.
If you don’t want to delete the custom field, but you don’t want it to be active for a while, you can turn it off instead. When you do this, all entity-related highlighting is removed from all tickets, just like when you delete a custom field.
To turn off entity detection for a custom field
- In Admin Center, click
AI in the sidebar, then select Intelligent triage > Entity.
- Click the entity you want to turn off.
- Under Update ticket fields with detected values select Don't update ticket fields.
- Deselect Highlight entity values in all messages.
- Click Save.
To delete a custom field associated with an entity
- In Admin Center, click
AI in the sidebar, then select Intelligent triage > Entity.
- Click the options menu icon (
) for the custom field associated with the entity you want to delete.
- Select Delete.
- Click Delete in the confirmation message.
Taking action on a highlighted entity in a ticket or messaging conversation
When the Highlight entity values in all messages option is selected, entities are highlighted in blue in tickets and messaging conversations. Agents can click a highlighted entity and quickly update the associated custom field with the highlighted value or else copy the value to the clipboard. This is applicable in most cases to values in drop-down fields.
When a highlighted value is completely shaded in blue, it means the associated entity custom field on the ticket has already been populated with that value. The only action available to the agent is Copy.
When a highlighted value is only outlined in blue (not shaded), it means the value hasn’t yet populated the entity custom field. The actions available to the agent are Update and Copy.
To take an action on a highlighted entity
- In Agent Workspace, open a ticket or messaging conversation.
- Click a highlighted entity value.
- Select from the following options:
- Update: (Available only if the highlighted value hasn't already populated the entity custom field) Updates the value of the associated custom field with the highlighted value.
- Copy: Copies the highlighted value to the clipboard.
16 comments
WhatsApp Connector
Does this functionality work with WhatsApp?
0
Ian Marston
When will we be able to update a custom field with the regex?
It would be very useful to be able to click and update the order number custom field
1
Andrew Chu
Hi team,
Can I have a quick check with you on whether Entity Detection has the capability to update automatically the other existing ticket fields already in place beside the newly created Entity field?
Is it something possible automatically or can only be achieved for now using conventional setup like triggers and automations?
0
Tim Sulzberger
Hi,
Firstly, I appreciate the new entity prediction feature.
I do have some feedback and questions about it. I've been trialling this, just for reporting purposes so far and with more frontline automation purposes later on.
My example is the prediction field I set up named “Product Line Prediction”, with a taxonomy of products which my company produces.
Questions:
Also a reporting recipe document would be useful. A tip I'll share with the community here regarding validating the AI, is :
using the Support tickets data set in explore reporting, add a row for Brand, ticket ID, Intent, [Your automatic Entity name}, [your manual categorisation field that staff use]. I've been comparing the prediction fields to the fields that staff use currently to categorise the ticket. This comparison helps isolate any unusual Intent or Entity predictions
0
barbara.graciano
Excelente esse recurso! Algumas dúvidas relacionadas a esse recurso:
a) Os campos personalizados criados no menu Objetos e regras > Tickets > Campos também poderão ser considerados como entidades, ou precisam ser novamente cadastrados um a um como previsões na Triagem inteligente?
b) No processo de inclusão de novas previsões, seria interessante que o menu Ações também tornasse possível a vinculação com campos personalizados existentes.
c) Se a detecção não se aplicar a campos personalizados tradicionais em uso na instância, e o administrador optar pelo uso de entidades, os relatórios configurados com campos personalizados tradicionais também terão um impacto, e precisarão ser atualizados com essas variáveis. Algum recurso mapeado para facilitar esse processo?
0
Ian Marston
Hi
Any update on the release on pattern detection yet?
Ability to identify the order number and populate a custom field would be a brilliant functionality for us.
It may seem trivial but shaving AHT by a few seconds on thousands of tickets and increasing accuracy would be great!
2
Enrique
Hello,
I am setting this up in our account but it doesn´t seem to work properly. The AI is simply reading whether the customer is saying the “magic” word indicated in the custom field, but not really paying attention to the context or anything.
Can anyone review this with me? thank you!
1
Elsa Aguilera
1264158222289 Any idea on when Regex fields or dates fields will be available to complete? Also, are you planning to include the possibility to create a definition of the information to extract from the ticket. (Exemple : Extract the date of the event the customer wants to attend / Result : Complete a date field on the form). Thanks
1
Erin O'Callaghan
Hi 1268286731109 and 1900994839044, regarding plans to add this functionality for the Regex field type, that enhancement is something the team is investigating for future implementation!
3
Andrew Lee
+1 to Yaniv, looking for the entity detection to be able to catch on synonyms or slight misspellings.
Example
- entity: Nikon D7500
- variation that should still classify with the same entity: nikoond7500, d7500, nd7500, etc.
2
Sign in to leave a comment.