Chat

Audience: Citizen Developers

Skill Prerequisites: Tokens

This action uses the OpenAI Chat API to get responses from AI.

Typical Use Cases

• Get AI-generated responses for chatbots
• Generate text content based on user input
• Implement AI-powered virtual assistants

Don't use it to

• Replace human-generated content entirely
• Process sensitive or private user data

Related Actions

None at the moment.

Input Parameter Reference

Parameter	Description	Supports Tokens	Default	Required
OpenAi API Key	The API key for Open AI.	Yes	empty string	Yes
Model	Choose from the following models: - gpt-3.5-turbo - gpt-4	Yes	empty string	Yes
Instructions	Provide instructions on how the AI should handle the messages. A typical example is to say "You are an helpful support assistant".	Yes	empty string	Yes
Previous Messages	If you are implementing a conversation, pass the previous messages as JSON, e.g. [{"role": "user", "content": "Hello!"}].	Yes	empty string	No
Message	The message that will get sent to OpenAi. It can be plain text, or a JSON conversation that follows this format [{"role": "user", "content": "Hello!"}].	Yes	empty string	Yes

Output Parameters Reference

Parameter	Description
Store Output Message	Stores the AI-generated response in the specified token.
Store Entire Conversation JSON	Stores the entire JSON of messages exchanged between the user and the AI in the specified token.
Store Entire Output	Stores the entire JSON response from OpenAI in the specified token.

Examples

1. Basic Chat

This example demonstrates how to use the Chat action to get a response from the AI. For example, you can import this JSON content into a workflow using the "Import Actions" button.

[
  {
    "Id": -1,
    "ActionErrorMessage": null,
    "ActionType": "OpenAi.Chat",
    "Condition": null,
    "Description": null,
    "IsDeleted": false,
    "EventName": "",
    "Parameters": {
      "ApiKey": null,
      "Model": {
        "IsExpression": false,
        "Expression": null,
        "Value": "gpt-3.5-turbo"
      },
      "Instructions": "You are an helpful support assistant.",
      "PreviousMessageJson": null,
      "Message": "How can I fix my printer?",
      "StoreMessage": "AIResponse",
      "StoreEntireConversationJson": null,
      "StoreJsonResponse": null
    }
  }
]

This will send a message to the OpenAI API asking for help with a printer problem. The API response will be stored in the AIResponse token.