Action Step
Empower you Agents to take Actions
Last updated
Empower you Agents to take Actions
Last updated
Action steps are how you empower you AI Agents to take actions when certain conditions are met.
An Action step can have multiple different actions which will be assessed and then taken if criteria are met. If an Action is taken, the data from its last step will be the output for the action step, and can be referenced in the main branch by reference the output from the Action step.
In each Action Step you can define,
The name of the step (i.e. Journal Actions).
The AI provider used to power this step.
The Actions available to this step.
In each Action you can define,
The name of the Action (i.e. Update Journal).
The condition of the Action.
This text determines when the action should be called (i.e. "A user has specified an edit to their journal").
The sensitivity of the Action (defaults to 80).
The Conversational Data required for this Action to be taken
Conversational Data is data that's processed from the messages exchanged between the AI and the User. Each piece of Conversational Data is make of up a
Name
This is used to reference the data later on in the Action branch.
Description
This is used to know how to extract the data from the conversation. A good description is required to ensure smooth functioning.
Conversational Data Examples
Name: Location, Description: The location the user has given in the conversation
Name: Mood, Description: The mood or current emotion the user has described in the conversation