Conversation AI (Action)

Written By Laudonia (Collaborator)

Updated at July 26th, 2024

The "Conversation AI" action allows you to ask questions, handle conditional responses, and customize the bot’s personality and instructions. This innovative feature empowers users to create dynamic, personalized interactions with their contacts, leveraging advanced AI technology. This article outlines the key highlights and functionalities of this feature, providing a comprehensive guide on how to utilize it effectively.

Key Highlights

  1. Interactive Questioning: Use this action to ask a specific question to a contact, integrating seamlessly with inbound messages.
  2. Conditional Branching: Design branches for conditional flow based on the contact's reply, allowing for tailored responses and actions.
  3. Customization: Personalize the bot’s personality and additional instructions to align with your brand and communication style.
  4. Bot Training Utilization: Leverage the existing training for the Conversation AI Bot, ensuring consistent and accurate interactions.

Setting Up the "Conversation AI" Action

Start by choosing the workflow trigger to initiate the workflow and configure the actions to precede the Conversation AI action to your preference. Next, select the "Conversation AI" action from the list of available workflow actions.

 

Prompt – Personality and Additional Instructions

Enable the "Advanced Bot Configuration" toggle to access and customize the bot’s "Personality" and "Additional Instructions." The prompt generated for the bot will combine:

  • The defined "Personality"
  • Any "Additional Instructions"
  • The main question for the contact
  • Context from the bot training conducted for the Conversation AI Bot
  • Previous conversations between the sub-account and the contact

 

Question for the Contact

The primary question that the bot will ask is set up here. This question is posed along with any reply to an inbound message from the contact, if applicable.

 

Time Out 

Specify the time frame in which the bot should wait for a response from the contact. When the contact fails to reply in the allotted time, the workflow automatically follows the Time Out branch, enabling you to implement follow-up actions or alternative steps to ensure continued engagement and effective workflow management.

 

Channel Selection

Select one communication channel from SMS, Facebook, or Instagram through which the bot will send messages to the contact.

 

Skip If Answered

Choose to skip asking a question if the answer already exists in the conversation history. Instead, the bot will bypass the question and direct the contact to the relevant branch or condition based on the existing response.

 

Branches and Conditions

Two standard branches are automatically created for each action, with the option to create additional branches as needed:

  • Time Out: Triggered when the contact does not reply within the specified time frame.
  • No Condition Met: Activated when the contact replies within the timeout period, but no specified condition is met.

 

Creating New Branches

Each new branch must have a unique name and a specific condition based on which the contact will be directed down that branch.

 

Once you have confirmed the configuration, select the “Save Action” button to add the action to your workflow. Remember to save the workflow to maintain your changes!

Mechanism of the Conversation AI Action

Once the "Conversation AI" action is configured, the following process unfolds when a contact reaches this action:

  1. Message Generation and Sending: The prompt is generated and the message is sent from the Conversation AI bot to the contact. If needed, context from the bot training in the Conversation AI Bot sub-account settings is utilized.
  2. Waiting for a Reply: The bot waits for a reply from the contact for the duration configured in the action.
  3. Time-Out Handling: If the contact fails to reply within the specified time frame, the process follows the "Time Out" branch, allowing for follow-up actions.
  4. Condition Checking: When the contact replies within the time frame, the bot evaluates which condition is met based on the reply. If no condition is met, the contact is directed to the "No Condition Met" branch, where further actions can be specified.
  5. Conditional Branching: If a condition is met, the contact proceeds down the corresponding branch, ensuring a personalized and relevant interaction.