Skip to main content

Overview of Execute Actions Tool Types for AI Agents in Whippy

Understand all available Execute Actions tool types and what each one allows your AI agent to do.

Maria Cairns avatar
Written by Maria Cairns
Updated this week

Why it matters

Execute Actions tools allow your AI agents to take structured actions inside Whippy or trigger external systems. Instead of only responding conversationally, agents can update records, send messages, manage sequences, and notify users.

Understanding each action type helps you choose the correct tool for your workflow and implement it properly.

Key Concepts

Action Tool: A tool created using the Execute Actions tool type that allows an agent to perform structured system actions such as sending messages, updating contacts, or triggering integrations.

Execute Actions Tool Type: A tool type that allows you to configure one or more actions inside a single tool. Multiple actions can run when the tool is triggered.

Action Type: The specific operation performed when the tool runs, such as sending a message or assigning a user.

Agent Instructions: The configuration where you define when a tool should be triggered and under what conditions. Creating a tool alone is not enough. The agent must have access to the tool and clear instructions on how to use it.

Step-by-Step: Create an Execute Actions Tool

  1. Navigate to Agents in the left-hand menu.

  2. Click Tools (jigsaw icon).

  3. Click Create Tool.

  4. Select Execute Actions from the Tool Type dropdown.

  5. Enter a tool name and save.

  6. Configure Basic Info and Tool Settings.

  7. Click Add Action to select and configure one or more action types.

After setup, open the relevant agent, enable the tool, and define in the Agent Instructions when it should run.

Available Action Types

Below is a summary of all available Execute Actions types and what each one does.

Send Message

Sends a message to the contact in the current conversation. Supports Static content or AI-generated content using Prompt mode.

Add Tags to Contact

Applies one or more existing tags to the contact record when executed.

Remove Tags from Contact

Removes one or more existing tags from the contact record.

Assign User

Assigns a specific user in your organization to the current conversation.

Unassign Conversation

Removes all currently assigned users from the conversation.

Opt Contact In

Updates the contact’s consent status to opted in, allowing outbound messaging according to your configuration.

Opt Contact Out

Updates the contact’s consent status to opted out, preventing further outbound messaging.

Notify Users

Sends an internal notification in Whippy to selected users.

Remove from Sequence

Removes the contact from one or more selected sequences, stopping future automated steps.

Add to Sequence

Enrolls the contact into a selected sequence so automated steps begin.

Move Sequence Step

Moves the contact to a specific step within a selected sequence.

Update Conversation Status

Changes the conversation status to Open or Closed.

Send Email

Sends an email to one or more recipients. Supports Static content or AI-generated content using Prompt mode.

Start Chat Agent

Activates a selected chat agent for the contact. Requires conversation_history in the receiving agent’s instructions if past messages are needed.

Send HTTP Request

Triggers an external API call using a GET or POST request to a specified URL.

Implementing Action Tools in Agents

Creating an Execute Actions tool does not automatically make it active.

To use any action tool:

  1. Open the relevant Agent.

  2. Enable the tool to give the agent access.

  3. Clearly define in the Agent Instructions:

    • When the tool should be triggered.

    • What conditions must be met.

    • Any required data collection or formatting.

Without explicit trigger logic in the instructions, the agent may not use the tool as expected.

Tips and Best Practices

  • Create separate tools for distinct workflows when logic differs.

  • Keep trigger conditions specific to avoid unintended execution.

  • Test each tool in a controlled conversation before deploying.

  • Combine multiple actions in a single tool when they should always run together.

  • Review tool configurations regularly as workflows evolve.

Troubleshooting

Issue

Possible Cause

Fix

Tool not triggering

Tool not assigned to agent

Enable the tool in the agent settings

Action running unexpectedly

Trigger logic too broad

Refine Agent Instructions

Incorrect action executed

Wrong action selected during setup

Review and edit the tool configuration

External integrations failing

Incorrect formatting or endpoint

Verify API configuration and parameters

Did this answer your question?