AI Assistant
| Component | Type | Description | |
|---|---|---|---|
 | AI Assistant | 🔀 action | Perform an action using LLMs |
The AI Assistant allows you to make a request to a selected LLM model to perform actions and provide an output. Examples use cases include:
- asking the AI to output a customized email body using placeholder values from a form
- asking the AI to extract text from uploaded images
- asking the AI to write custom code that is then inputted into the Run Code component
- asking the AI to output a value that is then used by the branch paths and conditional paths component to route the workflow in a certain way
Models​
The following models are currently available in the assistant:
- GPT 3.5 Turbo (Open AI)
- GPT 4 Turbo (Open AI)
- GPT 4 Turbo with Vision (Open AI)
Job description​
Provide a job description for the AI assistant. This helps it better understand it purpose and function and how it should handle both the inputs and generate the specified outputs.
Context​
Insert any text content to be used by the AI when performing its job. This can be text, placeholders or a combination of both. The AI will use this as context when performing its job and generating the outputs.
Image inputs​
For models with vision, you can use image URLs as an input. For example, if the user has uploaded an image from a form, you can use the placeholder for that question as an input.
Image inputs must be an image URL. You can pass in multiple image URLs, including a list of image URLs. If multiple image URLs are passed in, each image will be sent to the AI in the same request.
Setting the output​
Explanation​
By default, the AI can return an explanation for how it performed the requested action. Turn this switch on to output this explanation.
Setting outputs​
Describe the output to be returned by the AI. For each output, provide a description for the output and select the data type for the output.
