Automating tasks from Task Mining

  • Release version: Australia
  • Updated June 17, 2026
  • 4 minutes to read
    • Summarize
    Summarized using AI
    This content was generated using new OpenAI-powered functionality. Results are provided on an as is basis and are not guaranteed to be accurate or complete.

    Summary of Automating tasks from Task Mining

    ServiceNow's Automation Center integrates with Task Mining to streamline task automation by converting recorded repetitive tasks into AI-generated automations and AI agents. This process automates Excel-to-browser data entry tasks by decomposing recordings into modular steps, creating reusable AI agents that can be invoked via the Now Assist panel or scheduled with triggers. This reduces manual effort and technical complexity in automating desktop workflows.

    Show full answer Show less

    Key Features

    • AI-powered decomposition: Automation Center uses generative AI to break down task recordings into discrete UI and non-UI automations without manual scripting.
    • Support for Excel-to-browser scenarios: Currently focuses on automations involving copying data from Microsoft Excel into browser forms.
    • Automated agent creation: Generates AI agents in AI Agent Studio pre-loaded with instructions, input parameters, and AI Desktop Actions tools based on decomposed automations.
    • Review and refinement: Technical users can validate and refine AI-generated automations and UI element anchors before activating agents to ensure accuracy.
    • Reusable and accessible agents: Authorized users can invoke agents via Now Assist without needing access to Automation Center or AI Agent Studio.
    • Trigger-based automation: Agents can be configured to run automatically when specific triggers are met.

    Task Automation Workflow

    • Business users record repetitive tasks in Task Mining and submit automation requests.
    • Automation requests appear in Automation Center for technical review; no automatic notifications are sent.
    • Technical users use the Automations tab to generate and review AI-decomposed automations categorized as UI blocks (browser interactions) or non-UI blocks (background processes like Excel data reading).
    • Upon approval, an AI agent is created in AI Agent Studio, reviewed, adjusted if needed (especially UI anchors), tested, and then activated with assigned roles.
    • Authorized users can run the agent on Windows machines via Now Assist or schedule it based on triggers.

    Limitations and Requirements

    • Platform: Task recording and agent execution require Windows machines; configuration and agent creation can be done on any supported OS.
    • Supported Scenario: Only Excel-to-browser data entry tasks are supported currently.
    • AI Output Variability: AI-generated decompositions and instructions may vary and must be carefully reviewed and validated before activation.
    • UI Anchor Adjustments: UI elements identified in Task Mining may need manual correction in AI Desktop Actions to ensure reliable automation.
    • No automated notifications: Coordination between business and technical users is necessary as Automation Center does not notify users of new automation requests.
    • Plugin dependencies: Automation Center, AI Desktop Actions, and Now Assist for Automation Center plugins must be installed.
    • AI Skill dependencies: The user task step summarization skill must be activated to enable this feature.

    Practical Application for ServiceNow Customers

    ServiceNow customers can leverage this integration to accelerate task automation initiatives by converting Task Mining recordings directly into AI agents, minimizing manual development effort. By focusing on Excel-to-browser workflows, organizations can automate common data entry tasks efficiently. The process requires collaboration between business users capturing tasks and technical users refining and activating automations. Customers should prepare Windows environments for execution, install necessary plugins, activate required AI skills, and plan for validation to ensure accurate and reliable automation outcomes.

    Automation Center can receive task recordings from Task Mining, decompose them into discrete automations, and generate an AI agent in AI Agent Studio that executes those automations using AI Desktop Actions.

    About task automation

    Automation Center reads the task recording summary, breaks it into modular automations, and generates an AI agent — without requiring the user to manually author agent instructions or AI Desktop Actions tools from scratch.

    This feature currently supports only Excel-to-browser interactions, where data is copied from a Microsoft Excel file and it is entered into a form in a browser.

    Important:

    This feature uses generative AI to produce its output. AI-generated content, including decomposed automations and agent instructions, may be inaccurate or incomplete. Review and validate AI-generated output before activating an agent.

    Key benefits

    Automating tasks from Task Mining provides the following benefits:

    • Reduces the effort required to automate repetitive desktop work by generating agent instructions from recorded task data.
    • Automatically identifies and categorizes steps as UI-based or background interactions, reducing manual analysis by the technical user.
    • Technical users can refine AI-generated automations before creating an agent so the output reflects the actual task requirements.
    • Creates a reusable AI agent that any authorized user can invoke from the Now Assist panel without accessing Automation Center or AI Agent Studio directly. The agent can also automatically run if a trigger is set and it meets the trigger criteria.

    Task automation workflow

    The following graphic describes the end-to-end workflow for automating tasks from Task Mining:

    Figure 1. Automating tasks from Task Mining
    Workflow diagram showing the end-to-end process from Task Mining recording to AI agent invocation in the Now Assist panel
    1. A repetitive task in Task Mining is recorded (Business user). When the recording is complete, the user submits an automation request. The task timeline and a summary of user interactions are saved as an attachment to the automation request record.
    2. The automation request appears in Automation Center. A technical user reviews the request. There is no automated notification that informs the Automation Center user about the creation of the automation request in Task Mining.
    3. The automation request in Automation Center is opened (Technical user) and Generate automations is selected on the Automations tab. Automation Center reads the attachment and uses AI to decompose the task into discrete automations.
    4. Each decomposed automation is categorized as one of the following:
      • UI block: An on-screen interaction performed in a browser, such as entering data into a ServiceNow form. Automation Center generates a AI Desktop Actions UI block for these automations using screenshots from the task recording.
      • Non-UI block: A background interaction with an application that does not require on-screen control, such as reading data from a Microsoft Excel file. Automation Center creates a tool with instructions for the existing background application handler.
    5. The decomposed automations are reviewed (Technical user). If the output is unsatisfactory, the user can provide additional instructions and regenerate. Because the decomposition is AI-generated, results may vary between runs.
    6. When satisfied with the decomposed automations, the technical user selects Create agent. Automation Center sends the automation details to AI Agent Studio, which creates an agent pre-populated with instructions, a role description, input parameters, and AI Desktop Actions tools.
    7. The technical user reviews the generated agent in AI Agent Studio, corrects any inaccurate instructions, and tests the AI Desktop Actions tools. UI block anchors — the UI elements that AI Desktop Actions uses to identify where to click or type — may require manual adjustment because Task Mining and AI Desktop Actions capture anchor data differently.
    8. After testing and validation, the technical user activates the agent and assigns it to the appropriate roles. Authorized users can then invoke the agent from the Now Assist panel by describing the task they want to perform. You can also schedule the agent to run when some triggers are met. The agent will run the task as scheduled.

    Limitations and requirements

    Consider the following when using this feature:

    • Windows requirement: Task Mining recording and AI Desktop Actions agent execution require a Windows machine. Automation Center configuration and agent creation can be performed on any supported operating system.
    • Supported scenario: This feature currently supports only Excel-to-browser interactions, where data is copied from a Microsoft Excel file and it is entered into a form in a browser.
    • AI-generated output variability: Decomposed automations and agent instructions are generated by AI. Each run may produce different results. Review all AI-generated output before creating or activating an agent.
    • UI-block anchor adjustment: AI Desktop Actions UI blocks require anchors that identify the UI element to interact with. Anchors captured from Task Mining recordings may not match the format AI Desktop Actions expects and may require manual correction in the AI Desktop Actions editor.
    • No cross-product notification: Automation Center users don't receive a notification when a Task Mining user submits an automation request. The business user and the technical user must coordinate directly.
    • Plugin dependencies: Both Automation Center and AI Desktop Actions must be installed on the target instance before you can create an agent. Now Assist for Automation Center must be installed to use this feature.
    • AI skill dependencies: User task step summarization skill must be activated. For more information, see Activate skills for Now Assist for Automation Center.