Studio Web User Guide

DELIVERY:

Last updated Aug 7, 2025

Consuming API workflows

Consuming API workflows from Orchestrator

To run your published API workflows from Orchestrator:

In your Automation Cloud Orchestrator, go to the folder where the API workflow process exists.
Navigate to Automations > Processes.
For the desired API workflow process, select Start a job at the end of the coresponding row.
In the page that opens, provide:
- Execution Target options
- Runtime Arguments options. The Input section represents your API request input schema, and allows you to specify properties to use as inputs for execution.
To run your process, select Start.

Once your workflow has successfully run, you can view the generated response in the workflow execution details: Automations > Jobs.

Invoking API workflows in an Agent

Extend the capability of your UiPath Agents by integrating API workflows as tools. Agents automatically query these workflows to gather relevant context, enabling precise control over the data available to your agents and AI models.

The following scenario demonstrates how an API workflow can transform data retrieved from Workday, limiting the information exposed to agents to names and email addresses only. In the Workday sandbox, Betty Liu is configured as a manager with assigned direct reports and peers. Example queries for the agent include:

"Who reports to Betty Liu?"
"Provide all peers for Betty Liu."
"Do any of Betty Liu's peers have reports?"

Important: The agent does not store your Workday credentials, cannot directly query Workday, and relies solely on responses returned by the API workflow.

Before you begin, ensure the following prerequisites are met:

You published your Workday API workflow to Orchestrator. Confirm that the workflow runs successfully when executed from Orchestrator.
Platform units are assigned to your tenant.
To follow the provided Workday integration example, ensure your workflow includes:
- An input schema with fields for first name and last name.
- An output schema containing arrays for Peers and Direct Reports.

Create your agent manually. Alternatively, use the Autopilot generator with the following prompt:

"Query workday for manager, reports and peers details."

Your agent is created with Autopilot, which suggests the agent prompts and configuration. In this example, the suggested prompts were initially accepted, but the User Prompt was later simplified to contain only the {{ query }} variable.
In the Data manager panel of your agent, define an Input property called query, of type String. This ensures your User Prompt contains a valid reference.
Select Test on cloud. You are prompted to provide the query for your User Prompt. However, the Execution Trace panel shows that the agent currently has no access to Workday, so it cannot provide a response yet.
From the Tools section of your agent, select Add tool > API workflow.
Select the existing API workflow (the one that queries Workday), and provide a description for it. As a best practice, provide enough detail for the agent to automatically configure the workflow based on your prompt. A clear description typically includes details about your input parameters. For example:
"Takes a worker first and last name as input and provides their direct reports and peers in the response. The name input is the manager we will look up in Workday."
Run your agent again by selecting Test on cloud. Provide the following querry: "Who reports to Betty Liu?". The Execution Trace panel now identifies that the agent calls the API workflow, including:
- The request sent to the API workflow
- The response received from the API workflow
- How the agent uses the response to fullfil your query

Congratulations!

Now you can create variations of your agent user prompt. Depending on the query, the agent may trigger multiple calls to your API workflow before providing a complete response.

Using API workflows as tasks in Maestro

API workflows as tasks in Maestro encapsulate detailed API chaining and data transformations. This keeps the primary Maestro process clear, concise, and easy to follow. By isolating specific API interactions within API workflows, you maintain readability and promote reusability across multiple processes.

The following scenario demonstrates how Maestro calls Workday APIs and then sends a Slack team notification whenever an employee was terminated.

Before you begin, ensure the following prerequisites are met:

You published your Workday API workflow to Orchestrator. Confirm that the workflow runs successfully when executed from Orchestrator.

Ensure your API workflow includes:
- An input schema with fields for First Name and Last Name.
- An output schema containing arrays for Peers and Direct Reports.

Step 1: Creating an Agentic Process

Create an agentic process.
Add a Start event element. Name it "Employee Hired".
1. For the Action property of the Start event, select Wait for connector event.
2. For the Connector property, select Workday REST - Worker Created. This is the event that triggers the process.
Add a Task element and connect the Start event to it. Name it "Find Employee Relationships".
Add another Task element, name it "Send Notifications", and connect the Find Employee Relationships element to it.
Add an End event. Name it "Notification Sent".

This is the process setup. Whenever a new employee is hired, the process triggers the API workflow, which retrieves employee information such as their reporting structure and associated relationships.

Step 2: Configuring the Find Employee Relationship task

The API workflow required by the agentic process resides in the Find Employee Relationships task. Here is how to configure it:

Select the Find Employee Relationships task.
For the Action property , select Start and wait for API workflow.
For the API workflow property, select the published Workday API workflow. The properties panel displays inputs and outputs for the selected workflow. For this example:
- Inputs include Last Name and First Name.
- Outputs include details for Manager, Peers, and Reports.
Configure inputs as follows:
- First Name as a variable from the Start event response object. Open the Expression editor, select Insert variable and navigate to Employee Hired > response > Worker > Worker full name. Then add Split(' ')[0] to the expression. It should look like the following:
```
vars.response_1.worker.descriptor.Split(' ')[0]vars.response_1.worker.descriptor.Split(' ')[0]
```
  Where worker is the response object from the Start event response, and.Split(' ')[0] splits the full name by the space character and fetches the first part of the full name.
- Last Name as a variable from the Start event response object. Open the Expression editor, select Insert variable and navigate to Employee Hired > response > Worker > Worker full name. Then add Split(' ')[1] to the expression. It should look like the following:
```
vars.response_1.worker.descriptor.Split(' ')[1]vars.response_1.worker.descriptor.Split(' ')[1]
```
  Where worker is the response object from the Start event response, and.Split(' ')[1] splits the full name by the space character and fetches the second part of the full name.

Step 3: Customizing the Send Notification task

In this step, you configure the Send Notification task to use data returned from the API workflow executed at Step 2.

Select the Send Notification task.
For the Action property, select Execute connector event.
For the Connector property, select Slack, and configure it to use your Connection.
For the Activity property, select Send Message to Channel.
For the Channel name/ID, write the ID of the Slack channel where the notification should be sent.
For the Message property, select the output variables from Step 2: Manager, Peers, and Reports.
For the Send as property, select bot.