In this example, a Workflow is set up to do the following:
When a new ticket is created, the text in the Last Customer Comment in the ticket is analyzed using the Custom Prompt action to determine if the customer is providing feedback on a product. If so, this feedback is added as an Internal Note in the ticket.
For general information on setting up Workflows, refer to Configuring the Workflows for the Triggers+ChatGPT App.
In this Article
Creating the Workflow
Name and Trigger
Action that is Made by OpenAI ChatGPT
Zendesk Action
Enable the Workflow and Save
Opening the Trigger that is Created
Creating the Workflow
The following is a summary of how to set up the Workflow for this example.
Name and Trigger
Name | Enter the Name, for example, Product Feedback Categorize Ticket Based on Customer Comment |
Trigger |
This selects the Zendesk Trigger to use, and whether this is when the ticket is Created or Updated. You can select When a ticket is created (new Trigger) to create a new Trigger. You can also select an existing Trigger that has already been created for a Triggers+ChatGPT Workflow. For details on this, refer to Trigger in Configuring the Workflows for the Triggers+ChatGPT App. |
Action that is Made by OpenAI ChatGPT
The next fields determine what ChatGPT does. For this example, the following is added to the Workflow.
OpenAI ChatGPT Action | Select the OpenAI ChatGPT Action as Custom Prompt. |
GPT-4 or GPT-3.5 |
In order to properly use, or to test, Triggers+ChatGPT you need to have a paid OpenAI API account. This will allow you to select GPT-4. For further information, refer to Open API Key in Configuring the Workflows for the Triggers+ChatGPT App. |
OpenAI API Key |
The first time you set up a Workflow, you need to enter your OpenAI API Key which allows Triggers+ChatGPT to communicate with the OpenAI services. If you have already saved your key you will not need to enter this again. For details, refer to Open API Key in refer to Configuring the Workflows for the Triggers+ChatGPT App. |
User Message |
The User Message provides the instructions to ChatGPT. In the User Message you can use one of the following Placeholders to insert ticket data into the prompt. {{ticket.last_customer_comment}} In this example, the workflow is being set up to extract the feedback from the Last Customer Comment by including the Placeholder {{ticket.last_customer_comment}}. The instruction to ChatGPT is to read the comment, decide if there is feedback, and, if there is feedback, to summarize it. For this example, the User Message is added as follows. Determine if the customer is providing any feedback on our range of products which include plants and garden supplies. If the customer does provide feedback on such products please summarize the feedback. IMPORTANT: If there is no feedback provided about our products, respond with an empty text nothing else. This is the message from the customer. {{ticket.last_customer_comment}} |
System Instructions |
The System Instructions are optional and these help to set the behavior of the assistant for the ChatGPT action. You can use this to modify the personality of the assistant or provide specific instructions about how it should behave. For example, you could add this as "Act as a professional help desk support agent without adding any unnecessary comments." |
The following is how this example looks when setting up the Workflow.
Zendesk Action
Next, configure what the Zendesk Action is when it updates the ticket.
Add Comment to Ticket |
For this example, if the relevant text about Product Feedback is found in the Last Customer Comment, a Comment will be added to the ticket as an Internal Note. |
Enable the Workflow and Save
Enable the Workflow by clicking the Enable/Disable toggle and save it by clicking Save.
The Configuration displays and includes the saved Workflow.
You can open and edit the Workflow at any time from the Triggers+ChatGPT Configuration. If you want to check or make changes to the Trigger for the Workflow in your Zendesk Admin Center before enabling it, you can leave it as disabled, check the Trigger then open the Workflow and enable it.
Refer to Editing a Workflow in Configuring the Workflows for the Triggers+ChatGPT App.
Opening the Trigger that is Created
This example Workflow was configured selecting When a ticket is created (new Trigger). When the Workflow is saved, this automatically creates a Zendesk Trigger in your Zendesk Admin Center. You can open the trigger directly from the Workflow by clicking View Trigger.
Making Changes to the Zendesk Trigger
You can make changes to the Trigger, for example, to change the Conditions so that it is triggered either when a ticket is Created or Updated. In the Trigger you would change the following:
- Remove the Conditions in the Meet ALL of the following conditions section.
Ticket > Ticket Is Created - Add the following Conditions in the Meet ANY of the following conditions section.
Ticket > Ticket Is Created
Ticket > Ticket Is Updated
For an example of this change, refer to Making Changes to the Zendesk Trigger for the Categorize Workflow in Example Workflow - Categorize a Ticket based on the Text in the Customer Comment and Update the Categories as Tags in the Ticket.
If you make any changes, ensure that these are also relevant for any other Workflows that the Trigger is used in.
How the Workflow Updates the Feedback as an Internal Note in the Ticket
The following is an example of how this Workflow adds an Internal Note to the ticket with the feedback from the customer.
This is an example of the email that is sent by the customer.
The email arrives and creates a ticket.
A few moments after the ticket is created, the Workflow triggers and adds an Internal Note to the ticket with the feedback from the Customer Comment. In this example, the instructions in the Workflow are to summarize the feedback.
The Internal Note shows the name as the Zendesk user who authorized the Swifteq account to access the Zendesk account when the app was connected.