How to Create Different Types of AI Prompts


 

AI Prompt offers two distinct types of prompts that it can request based on the Request Type: 

  1. Human-Like Prompts
  2. Structured Data Prompts

 

1. Human-Like Prompts:
This type of request prompts a simple text response crafted in a human-like language. Below is an example demonstrating how to create an AI prompt for a human-like response:

  1. Navigate to AI Prompt Tab and Create AI Prompt: 

    Go to the AI Prompt tab and proceed to create a new AI prompt with the following properties:

    Request Type: Select "Human-like Text".
    Rest of the Properties: Customize them based on your preferences.


  2. Add AI Prompt Context:
    Click on Add button in AI Prompt Context Section:

    Include AI prompt context by selecting relevant fields.

    In this example, we add the "Subject" field of the Case object as prompt context.

  3. Generate Your AI Prompt:
    Generate and check your AI prompt that will be sent to AI Service, by passing a test record to the AI Prompt and clicking on "Generate" to view the generated prompt in the "Generated Prompt" sub-tab.

  4. Test Your AI Prompt:
    Verify the functionality of your AI prompt by clicking on "Test" and reviewing the response from the AI service in the "Response" sub-tab.

  5. Save Your Prompt:
    Ensure to save your prompt by clicking on the 'Save' button located in the top right corner of the page.
  6. Utilize Your Human-Like AI Prompt:

    6.1 Navigate to an Assignment Group and create a new Action Rule
    6.2 Select 'Update Field Using AI Prompt v1.0' as the Action Type. (Note that only AI prompts related to the selected context object type will be visible)

    6.3 Configure Trigger Conditions based on your scenario requirements
    6.4 Choose the AI Prompt to Execute

    6.5 Specify Destination Fields. 
    For example, in the scenario below, the response will be stored in the description field, while the raw response will be stored in a custom AI Raw Response field on the Case object.

    6.6 Click on 'Save' to save your action rule configuration.

2. Structured Data Prompts:

This type of request asks for a response in structured data format, enabling the returned values to be directly stored in fields with fixed data structures, such as number or boolean. Below is an example illustrating how to create an AI prompt for a structured data response:

  1. Navigate to AI Prompts Tab and Create AI Prompt: 
    Go to the AI Prompts tab and proceed to create a new AI prompt with the following properties:
    Request Type: Select "Structured Data".
    Rest of the Properties: Customize them based on your preferences.
  2. Add AI Prompt Context:
    Follow the same steps as mentioned in the Human-Like Prompts section.
  3. Add AI Prompt Structured Data Properties:
    Click on the "Add" button in the AI Prompt Structured Data section.

    Example 1: Add a structured data property that returns a boolean value (true or false).

    Example 2: Add a structured data property that returns a valid text value suitable for storage in a text field. Ensure the text conforms to the requirements of the field, such as plain text or pick-list values. If using pick-list values, you can select valid options as shown in the screenshot below.

    Click on "OK" to save each AI Prompt Structured Data Property as you create them one by one.
  4. Save your Prompt:
    Click on the "Save" button located in the top right corner to save your prompt.
  5. Generate and Test Your Prompt:
    Follow the same steps as in the Human-Like Prompt section mentioned above to generate and test your prompt.
  6. Utilize Your Structured Data AI Prompt:
    6.1 Navigate to an e2a email rule and create a new Field Assignment Rule.
    6.2 Select 'Update Field Using AI Prompt v1.0' as the Action Type (Note that only AI prompts related to the selected context object type will be visible)
    6.3 Configure Trigger Conditions based on your scenario requirements
    6.4 Choose the AI Prompt to Execute
    6.5 Specify Destination Fields. 
    In the example scenario below, when the prompt asks "Check priority", the response will be saved in the "test checkbox" checkbox field. Similarly, when prompted with "What are the properties?", the response will be stored in the "Description" field. Lastly, the raw response from AI will be stored in any field selected for "Raw Response Destination field", all fields on the Case object.

    6.6 Click on 'Save' to save your field assignment rule configuration.

By understanding these two types of responses, you can effectively craft AI prompts tailored to your specific use case, whether it involves generating human-like dialogue or structured data outputs.

 

Please contact us at support@ortooapps.com for any questions.

Was this information helpful?
Quick survey: How do you rate Email-to-anything?
★★★★★ - EXCELLENT
★★★★ - GOOD
★★★☆☆OK
★★☆☆☆POOR
☆☆☆☆ - RUBBISH

-