ChatGPT connector Thinklet

This page outlines parameters to use when configuring Thinklets with ChatGPT.

Connector: ChatGPT

The Nexthink workflow connector ChatGPT uses a predefined API connector that allows you to easily send requests to ChatGPT and receive responses that you can then directly use in the product.

There is no need to define the prompt format in the connector, as the connector combines two parameters—ChatGPT prompt text and ChatGPT prompt input— to build the ChatGPT prompt.

This solution is based on the ChatGPT API, which you can configure in the connector credentials section. For more information regarding ChatGPT API setup, refer to the OpenAI platform documentation.

Example use case:

You detect an operating system or application error code using a remote action or NQL. You then pass the information to ChatGPT to get suggestions for root cause and remedial actions. For this use case, configure the ChatGPT connector to use the following input parameters:

• For ChatGPT Model Name, input gpt-3.5-turbo. This can be set to any ChatGPT model you have access to.

• For Role, input user. Available roles depend on the selected ChatGPT model.

• For ChatGPT prompt text, input your custom ChatGPT request in plain text. For example, Please provide a brief explanation of the following error code and limit your response to less than 900 characters:

• For ChatGPT prompt input, input <application error code>.

Using these settings allows you to send the following request to ChatGPT: Please provide a brief explanation of the following error code and limit your response to less than 900 characters: DRIVER_IRQL_NOT_LESS_OR_EQUAL (d1)

The result of this prompt is available in the ChatGPT response output parameter of this connector. The output will provide you with a quick, concise response regarding the specific error in the prompt. You can then use this output elsewhere in your workflow, such as adding it as a work note to a ticket generated by the workflow.

• Parameters: Configure the data used as parameters for this action. Hover over the tooltip icon for more information on how the parameter is mapped and what data is required.

  • ChatGPT model name: ChatGPT model name, for example gpt-3.5-turbo.

  • Role: ChatGPT prompt role name, for example user.

  • ChatGPT prompt text: The text of your prompt to ChatGPT in plain text format.

  • ChatGPT prompt input: The input that appears after your prompt text which builds your full ChatGPT prompt.

• Outputs: View the outputs of the Connector Thinklet.

  • ChatGPT response: ChatGPT response in plain text format.

Refer to the Configuring Connector Thinklet documentation for more information about connectors Thinklets.