Create automation blocks from the automation request in Automation Center

---
title: Create automation blocks from the automation request in Automation Center
description: Create automation blocks \(on-screen tasks and background tasks\) using the automation request created in Task Mining. These automation blocks are used to create desktop actions and AI agent in Automation Center.
locale: en-US
canonical_url: https://www.servicenow.com/docs/r/better-together/generate-automations-tm.html
release: brazil
topic_type: task
last_updated: "2026-09-10"
reading_time_minutes: 4
keywords: [create agent, decomposed automations, automations blocks, desktop actions, automation center, UI block, non UI block, deterministic desktop actions]
breadcrumb: [Building desktop automations from Task Mining data, Solutions]
---

# Create automation blocks from the automation request in Automation Center

Create automation blocks \(on-screen tasks and background tasks\) using the automation request created in Task Mining. These automation blocks are used to create desktop actions and AI agent in Automation Center.

## Before you begin

-   Verify that Automation Center and AI Desktop Actions are installed.
-   Verify that ServiceNow Otto for Automation Center plugin is installed and the User task step summarization skill is activated.
-   Confirm that the business analyst submitted an automation request in Task Mining. For more information, see [Create an automation request in Task Mining](https://raw.githubusercontent.com/ServiceNow/ServiceNowDocs/brazil/markdown/better-together/create-automation-request-tm.md).

Role required: sn\_aia.admin, sn\_ac.automation\_admin or sn\_ac.automation\_technical\_user, and sn\_tm\_core.analyst

## About this task

When business analyst submits an automation request from Task Mining, the request appears in Automation Center with the task recording summary saved as an attachment. Use the **Generate automations** option in the Automation tab to break the automation into discrete automation blocks. These automation blocks are used to create an AI agent and desktop actions that are categorized as on-screen tasks and background tasks, which determine how AI agents run them in AI Desktop Actions.

-   **On-screen tasks**: Execute automation steps by interacting with the UI. These appear in the AI Agent Studio design workspace, where they can be tested, configured, and validated interactively before agent execution.
-   **Background tasks**: Process data based on preconfigured instructions \(such as reading information from Excel\) without requiring UI interaction. These desktop actions run in background and don't require interactive testing on the design workspace.

**Important:**

Because automations blocks are created using AI, results may vary between runs. Review all generated automations blocks before you proceed with creating an AI agent and desktop actions.

## Procedure

1.  Navigate to **Workspaces** > **Automation center Workspace**.

2.  Select the list icon \(\[Omitted image "autocenter-list-icon.png"\] Alt text: List icon.\).

3.  On the **Lists** tab, under **Build**, select **All Automation Requests**.

4.  Open the automation request submitted by the business analyst.

    **Important:** To find automation requests from Task Mining, check the **Intake source** column on the **All Automation requests** table and look for **Task mining**.

5.  Verify that the automation request contains the task recording attachment.

    The attachment contains the task timeline and interaction summary generated by Task Mining. Automation Center reads this attachment to generate automations. If the attachment is missing, ask the Task Mining user to resubmit the automation request.

6.  Select the Automations tab on the automation request page.

7.  Select **Generate automations**.

    The **Generate automations** window is displayed.

    \[Omitted image "unified-gen-auto.png"\] Alt text: Generate automations window

8.  Enter additional instructions in the **Optional context** field to guide the decomposition.

    Use additional instructions to focus or constrain the output. For example, you can specify steps to exclude or applications to prioritize. If you skip this step, Automation Center uses only the task recording attachment as an input.

    Automation Center reads the task recording attachment and uses AI to create on-screen and background automation blocks from the automation request.

9.  Review the generated automations.

    Each automation displays the following information:

<table id="table_gjv_zdc_rjc"><thead><tr><th>

Field

</th><th>

Description

</th></tr></thead><tbody><tr><td>

**Name**

</td><td>

AI-generated name for the automation.

</td></tr><tr><td>

**Execution Mode**

</td><td>

Provides information related to the type of automation block created.-   **On-screen tasks**: For On-screen tasks, screenshots are most important. Without screenshots, these automation blocks aren't generated.
-   **Background tasks**: For background tasks, instructions are most important.


</td></tr><tr><td>

**Description**

</td><td>

A summary of the steps the automation performs. This field is used as the basis for the AI Desktop Actions tool instructions and the agent instructions when you create the AI agent. Review this field carefully.

</td></tr></tbody>
</table>10. Open each automation to see the details.

    |Field|Description|
    |-----|-----------|
    |Automation name|AI-generated name for the automation.|
    |Target application|The application the automation interacts with, such as Microsoft Excel or a browser.|
    |Input and Output|AI-generated values for the data the automation receives and produces.|
    |Description|A summary of the steps the automation performs. This field is used as the basis for the AI Desktop Actions tool instructions and the agent instructions when you create the agent. Review this field carefully.|
    |Goal and intent|The primary goal of the automation.|
    |Guardrails|Considerations to run the automation smoothly.|

11. If the results are unsatisfactory, select **Generate automations** again.

    The **Generate automations** window is displayed.

    Provide the updates in the **Optional context** field. You can regenerate automations as many times as needed. Each run calls the AI model again and may produce different results, including changes to automations you did not intend to modify. Review all automations after each generation.

    **Important:** Each prompt starts fresh and doesn't carry forward the previous changes. To keep earlier changes, include all your requirements in one prompt.


## Example: Generating onboarding automation blocks

When the HR analyst's automation request arrives in Automation Center, the following automation blocks are created.

<table id="table_qth_wv1_hkc"><thead><tr><th>

Task type

</th><th>

Automation blocks

</th></tr></thead><tbody><tr><td>

On-screen tasks

</td><td>

-   **Fill HR form**

Fill HR system form with new hire data.

Requires screenshots and UI interaction metadata.

-   **Add employee to Outlook**

Add employee to Outlook distribution list.

Requires browser navigation and form interaction.


</td></tr><tr><td>

Background tasks

</td><td>

-   **Read data from Excel**

Read new hire information from Excel file.

Operates silently based on instructions.


</td></tr></tbody>
</table>The technical user reviews the generated blocks:

-   Verifies that the on-screen form-filling task captured all required fields
-   Confirms the distribution list task navigates to the correct team list
-   Reviews the background task instructions for reading Excel data correctly

## What to do next

You can now create an AI agent and desktop actions from the generated automation blocks. For more information, see [Create an AI agent and desktop actions from Automation Center](https://raw.githubusercontent.com/ServiceNow/ServiceNowDocs/brazil/markdown/better-together/create-agent-automations.md).