AIP Generated Content

The AIP Generated Content widget can display the response from an AIP Logic function or visually stream the response from an LLM (similar to the Stream LLM response into variable event).

Configuration options

The AIP Generated Content widget has the following three options for generating a response:

1. Logic: This option requires selecting an AIP Logic function.

2. Direct to LLM: This option enables displaying the response of an LLM in real time within the widget. It has the following configuration options:

   • Prompt: The string variable of your prompt.
   • Clear output when prompt changes: When enabled, this will clear the AIP Generated Content widget output when the prompt variable changes. This is useful when the user is directly interacting with the prompt.
   • Temperature: The temperature to use with the model, a number between 0 and 1. Higher values, like 0.8, will make the output more random, while lower values, like 0.2, will make the output more focused and deterministic.
   • Model: The language model to use. Four OpenAI models are supported: GPT-3, GPT-4, GPT-4 32K, and GPT-4-Turbo.

3. LLM via prompt function: This option is the same as Direct to LLM, but the prompt is a function instead of a string.

Other configuration options

• Output variable: The string or object set variable that stores the response.
• Icon: The icon to display at the top of the widget; defaults to Palantir's AIP icon.
• Button label: The label to use within the button; defaults to "Generate result using AIP".
• Title: The title to display in the center of the widget.
• Loading state message: The message to display as the response is loading.
• Show loading spinner: When enabled, it shows a spinner when loading instead of the loading state message.

The following example shows what the AIP Generated Content widget looks like when displaying a string response: