Getting Started with Triage Quickstart

What is it?

Triage Quickstart is an extension of Triage that empowers you with a mix of pre-built and custom models that automatically suggest field values on cases. With just a few clicks, you can easily turn on the pre-built models directly in your dashboard without needing to work with the Forethought team. For custom models, a paid plan is required. See “How can I use custom models?” in the FAQ section below for more information.

Why is this important?

We know that support agents are focused on getting to customer issues as quickly as possible. Unfortunately, they’re stuck doing mundane tasks like manually prioritizing, categorizing and routing cases. For customers that are already using AI to triage tickets, there’s typically manual effort required to get it up and running (e.g. coding).

With Triage Quickstart, customers can automate the process of tagging cases with little to no effort involved!

What tags can I start adding to cases?

Triage currently provides four pre-built models:

1. Sentiment Detection

Sentiment Analysis detects patterns in text and automatically classifies a customer's overall emotions as positive, negative, neutral or mixed.

• Positive – The text expresses an overall positive sentiment
• Negative – The text expresses an overall negative sentiment
• Mixed – The text expresses both positive and negative sentiments
• Neutral – The text does not express either positive or negative sentiments

2. Spam Detection

Spam cases clog queues and pose a threat to the teams. This classifies text/messages as spam or not spam.

• Spam
• Not Spam

3. Language Detection

Classify tickets by language and route them to the appropriate team member.

• English
• French
• German
• Italian
• Korean
• +172 more languages

4. Sentiment Intensity Detection

This detects patterns in text and automatically classifies the intensity of the customer's intent.

• Highly negative - Indicates strong discontent or frustration. Customers feel extremely dissatisfied, often expressing significant issues or disappointments.
  Example:
  Customer: "This is the worst product I've ever used. It broke within a day, and customer service was rude and unhelpful. I regret buying it, and I'll never purchase from this company again."
• Moderately negative - Reflects general dissatisfaction. The sentiment is negative but not as intense as 'highly negative.' Customers may express frustration but may also show a willingness to resolve the issue.
  Example:
  Customer: "The product doesn't work as well as I expected. It has some noticeable flaws, and while it gets the job done, I'm pretty disappointed overall. I wouldn't recommend it unless there are no other options."
• Neutral - Indicates indifference or a balanced view. The customer may be neither particularly satisfied nor dissatisfied, often presenting facts or neutral feedback.
  Example:
  Customer: "The product is okay. It does what it's supposed to, but there's nothing special about it. It neither exceeded nor fell below my expectations, just average."
• Moderately positive - Represents a satisfied or pleased sentiment. Customers express approval or appreciation, often highlighting good experiences or service.
  Example:
  Customer: "I'm really happy with this purchase. The product works well, and while there are a few minor things that could be improved, it definitely met most of my expectations. Overall, a good experience."
• Highly positive - Reflects extreme satisfaction or joy. Customers express strong approval or delight, often sharing enthusiastic praise or admiration.
  Example:
  Customer: "This product has completely transformed my life! The quality is exceptional, the customer service is outstanding, and I can't recommend it enough. I'm beyond satisfied and will definitely be a loyal customer for years to come."

If needed, you can work with the Forethought team to build custom models to meet your classification needs. Some examples include:

• Root cause analysis: Suggests root cause of customer inquiry.
• Product area classification: Suggests product inquired about
• Case reason: Suggests reason customer submitted case (like password reset) 
• Escalation: Prioritizes cases 
• and much more!

How do I get started?

For prospective customers, please start with step 1. For existing customers, continue to step 3.

1. Fill out our free trial form with your information.
   

   

2. Once completed, you should receive an email with login details.
3. Log into the Forethought dashboard and click Models under Triage. You will see the four pre-built models.
   

   

4. Click on the model you'd like to set up.
5. Click on Configure Integration to navigate to the Integrations page, where you can set up your configurations. Learn more about Forethought integrations.
   

   

   Learn more about Forethought integrations
6. Go back to the Models page and select a model you want to set up.
7. Return to the Models page and select the model you wish to set up. Then, click Get Started.
8. Configure your connected help desk to meet your specific needs.
   • If you have multiple connected help desks, use the dropdown menu to select the one you want to configure.
   • The output field will automatically populate with a suggested name for this model. You can modify the name, but be aware that if a field with that name already exists, it will be overwritten.
   • Forethought will automatically create a new ticket field using the field name set in the Output field box. If the field already exists, it will be overwritten.
9. Click Save.
10. Test the model on the right side of your screen by inputting a phrase or a sentence that a customer might say. You can also copy and paste the contents of your ticket (subject and description).
    • For example, to test the Sentiment Intensity Detection model, you might input:
      "Hey, I just got my order, and it doesn’t look like the picture on your website."
    • Then, click Classify text. The Triage LLM will classify your input.
      

      

11. Click the toggle to activate your model.
12. Once your model is active, all incoming tickets submitted to your help desk via email and web forms will be tagged by the activated model. The predicted values for each activated model will be recorded in the designated output field. 

Frequently Asked Questions (FAQs)

Q: How can I test a model before activating?

A: Triage Quickstart allows you to test ticket classification in each model.

Copy or type your ticket contents (Subject and Description) in the right-hand sidebar, then click Classify text.

Spam Detection Model

Sentiment Intensity Detection Model

Language Detection Model

Q: How can I use custom models?

A: Custom models are not available as part of a Triage Quickstart trial. To enable custom models, please click Upgrade Plan, and a member of the Forethought team will contact you with more information.

Q: How does Triage work on the backend?

A: When a new ticket is created in your Zendesk, a trigger fires, notifying Forethought servers via a webhook. This effectively requests a prediction from Triage, and after classifiers have determined tags for this ticket, the specified output fields on the ticket will be filled in via an API call.

Q: I filled out the form but haven’t received my login details. What should I do?

A: If you have any trouble getting access to the free trial, please reach out to support@forethought.ai.

Q: What is the duration of the trial?

A: There is no time limit, but we do limit the number of predictions made to 5,000. You can see your usage on the bottom left of your dashboard. Once you hit your free trial usage, the predictions will automatically stop.

Q: What if I only want Triage to tag certain tickets?

A: If you want to make sure Triage does not tag certain tickets, you can add conditions to the Zendesk trigger that notifies Forethought of newly created tickets. For example, if you only want to tag tickets coming from the “Email” channel, you can add a trigger condition of “ticket: channel is Email”.

Q: Once a model is activated, how will agents know it’s working?

A: After your admin activates the model, agents should see an additional field with the appropriate tag at the help desk. 

Q: What help desks are supported for Triage Quickstart?

A: Currently, only Zendesk and Salesforce are supported. Support for additional help desks is on the way!

Q: What kinds of tickets are supported?

A: Currently, we support all tickets submitted via email and web form.

Q: What value is written to the output field in my helpdesk?

A: The output with the highest confidence score is written to the helpdesk. For example, if the language detection model is used and the output is the following: 

Spanish is written to the output field since it has the highest confidence score.

Q: Can I pause/deactivate my trial and reactivate it later?

A: Yes, simply deactivate your active models and reactivate them as you please.

Q: What should I do if I want to upgrade to the paid plan?

A: To upgrade, click Upgrade Plan in the Forethought dashboard (under Triage > Models) and a member of the Forethought team will contact you with more information.