Register the Human Approval Integration as an Agentic AI Tool

An integration can become an agentic AI tool for AI agents. AI agents can then invoke the integration as a tool to accomplish a specific task.

Let's register the human approval integration as an agentic AI tool so that it can be used by the AI agent.

Prerequisites:

Create and Activate the Human Approval Integration
  1. In the navigation pane, select Projects.
  2. Select the project in which your integration is located. For the tutorial, the project is Simple Expense Approval Project.
  3. In the Integrations section, find the integration that you want to register as a tool: Request Human Approval.

    Simple Expense Approval Project dashboard in Design view. The Integrations section lists four integrations—Request Human Approval (active), ReAct Pattern Intg, Get Tool Data, and Auto Approve Expense Report (all configured)—showing their version and status. The Connections section lists four connections—REST_TRIGGER_FOR_HUMAN_APPROVAL, LLM_Connection, REST Connection, and REST_TRIGGER_FOR_AUTO_APPROVAL—with their types (Trigger or Invoke) and configured status. The top bar includes a search box, view options, and action icons. A vertical navigation menu appears on the left side

  4. Check that your integration is Active. If it's not active, activate it by clicking Actions Actions icon, and selecting Activate.
  5. Register the integration as a tool.
    1. Next to Request Human Approval, select Actions Actions icon, and select Create agentic AI tool.

    The Create Tool panel is displayed.

  6. Enter information for the tool.

    Create tool screen for adding a new agentic AI tool. The form displays fields for Type (Integration), Project (SIMPLE_EXPENSE_APPROVAL_PROJECT), Integration (Request Human Approval 1.0), Name (Request Human Approval), Identifier (REQUEST_HUMAN_APPROVAL), and Description (“Requests human approval for submitted expenses”). Cancel and Create buttons are at the bottom.

    Field Description
    Name

    Required. Automatically populated from the integration name.

    Not sent to the Large Language Model (LLM).

    You cannot change the name after the tool has been created.

    Identifier

    Required. Automatically populated from the tool name.

    Uniquely identifies the tool in the project.

    Sent to the LLM as tool metadata. The AI agent uses this information to understand the purpose of the tool and when to use it.

    You cannot change the identifier after the tool has been created

    Description

    Required. Automatically populated from the integration description, if available.

    Sent to the LLM as part of the system prompt.

    The description helps the LLM decide when to use the tool. Add information to clearly describe what the tool does and when to use it. Clear descriptions help LLMs use tools correctly and reduce errors.

    Enter the following description: 

    Requests human approval for submitted expenses.
  7. Click Create.

    The tool details page is displayed. The tool details page lists the tool description, guidelines, and input parameters. The tool description and guidelines are sent to the LLM. Guidelines are constraints to limit tool behavior and respect corporate policies. Guidelines influence the LLM decision process.

  8. Review tool parameters.

    When we take a look at the Parameters configuration, we see integration input parameters. The claimamount parameter is listed and it is sent to the LLM because it's marked as Visible. We don't need to make any changes.

    If you were working with an integration that had many parameters, you could make visible to the LLM only parameters relevant to the tool.


    Request Human Approval tool details page showing a description that reads "Requests human approval for submitted expenses." The guidelines section is empty. Below, the Parameters configuration table lists one parameter, "claimamount," which is visible, required, and of type string. Other columns for constant, default, enum values, and description are empty.

  9. Click Save to save your changes.
  10. Click Back Back icon to return to the Project page.

    The integration can now be used as a tool in an AI agent.

Next step: Since we already created an AI agent in the first tutorial Tutorial: Build and Run Your First AI Agent in the step Build the AI Agent, let's add the human approval tool to the AI agent. Go to Add the Human Approval Tool to the AI Agent.