This feature allows you to set up and use workflows in Teams on files. You can create a flow on a file in the overflow menu of the Workflows app. Improve your productivity by using one of the templates that are available to be done to a file, or by using AI to create your flow from scratch.

Prior to this feature, you needed to write complex expressions in your cloud flows in Power Automate to process data, do conversions, manipulation, or comparison. With this feature, you can create complex expressions or edit your existing expressions simply by submitting natural language prompts to Copilot in Power Automate.

You need Copilot access in your tenant to use this feature. Turning off Copilot turns off this feature, too.

Generative actions use generative AI reason over unstructured content and determine what steps and sequences it needs to achieve a process outcome based on instructions and parameters provided by you in natural language.

Building a generative action within cloud flows involves the following steps:

• Describe the objective of your process in natural language, and let AI develop an automation plan to achieve it.
• Refine the automation plan generated by the large language model (LLM) and adjust inputs, outputs, and variables as needed.
• Add reference sources and guidelines in natural language that influence the LLM as it executes the flow.
• Validate the intelligent action by reviewing how it is achieving the business objective prior to moving the flow into production.
• Review production run history, analytics, and individual flow run details to ensure your process is performing to expectations.

The ability of generative actions to easily incorporate both structured and unstructured data makes it flexible as information resources are updated. Its reasoning abilities make generative actions uniquely beneficial in the building and maintenance of complex process flows to handle conditions, exception management, and process loops.

View and manage all flow dependencies involved in your multi-flow process. Parent/child flows, desktop flows, and work queues are visualized through the process map management experience.

End-to-end observability through a process-centric view of all automation logs allows you to do the following:

• Easily perform impact analysis of failing sub-flows.
• Efficiently navigate between flow run logs, credentials, and connections.

Following is a summary of the savings feature:

• You can define a saving rule on a cloud flow, on a work queue, or on a desktop flow.
• The rule initially defines how much time and money each of their child-item generates (child-item: cloud flow run, work queue item, desktop flow run).
• The savings is aggregated and displayed in the automation center and in the new Value Hub in Power Platform admin center.

The new designer automatically saves a copy of the flow to browser storage upon a failed save, even with errors. This capability comes in handy on two occasions: 1) When the underlying service is going through an outage and when makers need to avoid losing their unsaved changes by exiting out of their flows, and 2) For non-solution flows, which lack the 'Save draft' functionality, you can exit out of your flow with errors and come back at a later time to fix the errors and save the flow.

An information banner appears on the designer notifying you when the designer is able to save the flow copy to the browser's storage. You can now exit your flow.

Upon revisiting the flow on the designer, the previously saved version loads on the designer by default. The unsaved copy is recoverable through the 'Recover' button on the info banner.

You can now fix errors on this copy of the flow. If you don't save this copy, the previously saved version of the flow is still accessible upon reloading the tab. However, saving this copy overrides the previously saved version of the flow. It also clears the browser storage, as there are no unsaved changes to the flow.

When you select the plus sign (+) for an action in the designer, the action pane opens. The action pane lets you search and browse for connectors and actions. Today, it's difficult to find the connector makers want, as there's no way to access favorite or highly frequently used connectors, or most common built-in tools of a flow. With this change, makers can favorite any connector or action so they show up in the Add an action pane by default under the Favorites section. They can quickly access AI capabilities in the product under the AI capabilities section, and they can quickly leverage building blocks or built-in tools like Control, Data operation, JSON, and more, under the Builtin tools section.

This improvement to the Browse connectors action pane in cloud flows makes it more accessible and efficient to makers while adding artifacts to their flow.

Prior to this feature enhancement, if you needed to filter out data that your 'Get items', 'Get files', or 'List rows' SharePoint and Excel actions by a certain rule (column name, and others), you needed to write an OData query yourself. With this enhancement, you can use a simplified OData editor, which doesn't require you to write complex OData queries yourself. You can use the visual editor to build your OData query.

This feature is automatically turned on for makers.

To use this feature, you don't need to know technical aspects. When you want to add an action, examinate the property value of multiple steps at design time, or the output values of multiple steps at runtime simultaneously. All that information displays inline by default.

Without inline view, you must select each card individually to display only the value of this certain step on the side pane. Then you must select the next card to display the value on the side pane again, without the ability to display and compare multiple values at once.

To do so from Power Automate for desktop, open the console, navigate to the My flows tab, and locate the flow that you want to create a description. Then, go to flow properties and under the general flow properties tab, press the Let Copilot create description button. Copilot will then analyze your flow and generate the flow description.

You can also generate a flow description from the portal. To do so, locate the desktop flow that you want to create the description for and press on the details item from the flow menu. Then, press on the edit button in the details section of the flow and finally select the Let Copilot create description button.

This feature is automatically turned on for makers with access to Copilot.

You don't need to know technical aspects to use this feature. When you want to add an action at a certain place, you see an option to describe to inline Copilot what actions you want in natural language. Then, inline Copilot creates these actions for you automatically.

Without contextual Copilot, you must describe to Copilot where exactly you want those new steps being added to in the chat window, which can be very time consuming.

With Copilot's generative answers functionality, you can easily find the information you need to create and manage your flows and any other aspect of Power Automate for desktop. The generative answers are integrated into the Power Automate for desktop UI in both the console and the designer, making it easy to access and use. Just type your question in natural language, and Copilot provides you with the most relevant information from our documentation. To use it, select the copilot icon in the console or designer UI and start getting answers to your questions.

This feature is automatically turned on for makers who have access to Copilot.

You don't need to know technical aspects to use this feature. You see the auto-generated summary at the top of the page. You can expand it to read more details or select the download button to download it as a PDF file.

This feature saves you time by automatically creating flow documentation for members without access to the flow. Also, other members might not know how to read a raw flow definition.

The next best action suggestion feature has been designed to streamline the automation process, making it easier for you to create and automate workflows without requiring extensive flow creation knowledge.

With this new addition, you will be able to create more complex workflows with ease. The feature will suggest the next action that makes sense based on the context of the current workflow and the actions that have already been added.

Natural language to Code is a new AI capability that we are adding in Power Automate for desktop that allows you to quickly generate code that is used in the scripting actions by just describing it. This feature is available in the "Run PowerShell," "Run VBScript," "Run DOS command," "Run Python," and "Run JavaScript" actions. You can simply describe what you want to achieve, and the tool will generate the script for you. This feature makes it easy for users who are not familiar with scripting languages to automate their tasks.

Troubleshooting Copilot can assist users with a new feature that can rectify errors in your cloud flow. These corrections are based on issues that arise during authoring and debugging and can be implemented to amend your flow within the Power Automate designer.

Administrators, CoE teams, business users and makers with access to flow run histories can query past runs in natural language across their environment. They can ask, for example, how many flows ran yesterday on which machines or groups, what did the success rate look like, and how do they compare to last month. Ask questions about recent runs, their durations, volumes, failures, failure frequency, machines, machine groups, parent flows and much more.

The Shadow DOM support in Power Automate for desktop introduces advanced functionality to handle encapsulated DOM structures effectively. Key features include:

• Enhanced element detection: Accurately locate and interact with elements within shadow roots using advanced selectors.
• Dynamic Shadow DOM handling: Automate dynamic or runtime-generated Shadow DOM components without interruptions.
• Nested Shadow DOM support: Traverse and interact with deeply nested shadow trees.
• Cross-domain iframe compatibility: Handle Shadow DOM elements embedded in cross-origin iframes for seamless automation.
• Recorder and playback updates: Effortlessly capture interactions inside Shadow DOMs and replay them accurately.

Record with Copilot in Power Automate for desktop allows you to build desktop automations by sharing your screen and describing the task you want to automate—just as if you were explaining it to someone. During the recording, record with Copilot captures your voice, mouse, and keyboard inputs. It then processes and converts these inputs into a desktop flow that you can review, edit, and save before running. This makes automation accessible to everyone, regardless of technical expertise, and significantly speeds up the process of creating desktop flows.

VM (virtual machine) image capture allows you to reuse the custom VM image for provisioning other hosted machines or hosted machine groups. It performs the following:

• Snapshot: Point-in-time capture of the VM's disk.
• Generalization: Removes machine and user specific information from the VM so that it can be reused.
• Storage in your Azure Compute Gallery: Allows the VM image to be stored and usable by Power Automate services.

Work queues in Power Automate can play a crucial role in improving the efficiency, scalability, and resiliency of automation and help prioritize work.

This is how you can configure Service Level Agreements (SLAs) for work queues:

• Type of SLA
  • Time to live (TTL)
  • Business hours
• SLA engine
  • Prioritize work queue items that are nearest to the SLA deadline.
  • Identify work queue items at risk.
  • Monitor the state of SLAs for work queues.

This feature provides support for joining hosted machine group bots to both on-premises Active Directory (AD) and Microsoft Entra ID when using custom virtual networks (VNETs).

• A custom virtual network (VNET) is configured using Azure Virtual Network. Access to the domain controller is required from the custom VNET.
  • A custom VNET allows you to access on-premises resources when running your automation.
• Microsoft Entra hybrid join is configured as part of Power Automate network connection setup.
  • Microsoft Entra hybrid allows your hosted machine group bots to enroll in both your on-premises Active Directory (AD) and Microsoft Entra ID.

Copilot in Power Automate for desktop offers an interface that simplifies the automation creation process. With natural language and Power Fx syntax capabilities, you can create desktop flows by providing instructions in natural English language or using Power Fx syntax. This eliminates the need for algorithmic or pseudocode knowledge. With Copilot, you can generate automations with ease, saving valuable time and resources.

The new Copilot side panel feature in Power Automate for desktop enhances user experience by providing detailed explanations of automation flows and subflows. Users can easily navigate to the Copilot side panel, select the appropriate skill from the skill picker, and ask questions like "What does the subflow do?" or "What do actions from 3-12 do?". This feature delivers clear insights into automation processes, streamlining troubleshooting and maintenance of complex automation. This empowers teams to quickly identify and resolve issues, ensuring consistent operational efficiency and SLA compliance.

In the Power Automate for desktop console, you can switch between the default color mode and dark mode. This setting allows the console and flow designer to use a dark theme based on your preference. The functionality is available in the console settings.

Endpoint filtering in Power Automate for desktop offers an added layer of control and security for your automation tasks. As an admin, apply endpoint filtering using the existing Data Loss Prevention (DLP) feature in Power Automate. Create an allow list of applications and websites that your automation scripts can interact with, or a deny list of those that should be restricted.

This feature targets UI and Browser automation actions, ensuring that your automation adheres to your organization's policies. Endpoint filtering is checked the same way DLP policies are currently verified in Power Automate for desktop. Leverage endpoint filtering to tightly manage which endpoints are accessible, enhancing the precision and security of your automation workflows.

• Desktop flows have versions. Each time you open a flow, you can make changes and save drafts until you decide to publish it.
• Version history in the flow designer for Power Automate desktop shows who made the changes and when.
• Compare changes between the current version and a previous version or between any two versions you select.
• In the console for Power Automate desktop, the context menu has a version history option. This shows version information in the properties window of the desktop flow.
• Some of this functionality will also be available in the portal.

The optimize desktop flows with intelligent insights and recommendations feature in Power Automate for desktop is designed to provide users with proactive, intelligent recommendations during the creation of desktop flows. This feature leverages historical flow data, usage patterns, runtime performance, error profiles, and established best practices to offer tailored, context-sensitive insights that can significantly enhance the efficiency and reliability of your automations.

In the Error Avoidance category, this feature helps prevent run-time errors by alerting users to potential issues before they occur. For example, it can warn of data loss prevention (DLP) policy violations that could cause a flow to fail, and it offers workarounds where possible. It also identifies elements that violate flow checker policies and suggests automatic remediations. Furthermore, it detects previous selector failures and self-healing events, recommending users to review or remediate these selectors accordingly. This feature also highlights potential flow timeouts by analyzing past run history, and it offers code-scanning recommendations to avoid pitfalls like infinite loops or uninitialized instances.

In the Error Remediation category, this feature assists in fixing design-time errors that might otherwise go unnoticed. It provides suggestions to complete any missing action properties, including variables, and helps correct syntax errors in Robin or Power Fx. This guidance ensures that flows are logically sound and free from common errors, making the development process smoother and more efficient.

The Code Optimization category focuses on improving the performance and maintainability of desktop flows. This feature suggests more performant or understandable actions, identifies logical shortcomings, and proposes possible solutions. It recommends wrapping actions in try & catch blocks to handle exceptions gracefully. It adds comments or regions for better readability, and breaks large subflows into smaller, more manageable units. These optimizations enhance flow performance and make the code easier to understand and maintain. The AI-powered Copilot facilitates these improvements by providing clear remediation steps and linking to relevant documentation. This empowers you to implement best practices with ease and confidence.

You can create and assign tags to your desktop flows through the console. This feature allows you to search and filter your flows based on tags, making it easier to manage and locate specific flows.

Tags are centrally managed in the Dataverse environment and are not limited to desktop flows. This means that other Power Platform assets can also use this functionality in the future.

Copilot in Power Automate for desktop now supports contextual memory, allowing you to build flows by referencing previously provided prompts. This eliminates repetitive input, making automation creation more intuitive and efficient.

Key Features and Benefits:

• Contextual Understanding: Copilot retains details from earlier prompts, such as file paths, parameters, or objects, for seamless reuse.
• Efficiency: Simplifies complex workflows by removing the need to retype information across multiple steps.
• Improved Accuracy: Ensures consistency and minimizes errors by accurately referencing previous inputs.
• Intuitive User Experience: Facilitates smoother, smarter automation creation with minimal effort.

A new action named "Get Credential" in Power Automate for the desktop will allow you to securely retrieve and utilize credentials directly within your desktop flows. During the preview phase, this action will support credentials created via CyberArk, accessible within the specified section of the Power Automate portal. This feature ensures that credentials are handled securely, aligning with best practices for data protection.

Once allowed by users, the feature uses AI to locate elements on the screen at the execution time. The repaired selector is used in future executions. Plus, it's optional and can be disabled if you prefer.

• Self-healing refers to the case that a UI element cannot be found at the execution time.
• Self-healing can be applied only for UI and web automation scenarios, and subsequently for Desktop and web UI elements.
• With self-healing, Power Automate for desktop can detect UI elements on the screen.
• When a UI element and its selector are retrieved, the user must confirm to heal and save.
• This feature can be disabled by the user.

The resolution setting for unattended runs in Power Automate for desktop allows you to configure screen resolution and DPI settings directly within each flow's properties. Define and apply precise resolution settings that match your automation's requirements.

Key Capabilities:

• Per-flow configuration: Set specific resolution and DPI preferences at the automation level, ensuring compatibility regardless of the machine executing the task.
• User-Friendly interface: Access resolution and DPI settings through an intuitive interface, reducing the complexity of traditional configuration methods.
• Persistent settings: These configurations persist through software updates and across machines, eliminating the need for repetitive adjustments.

This feature allows you to share desktop flow connections with a Pipeline SPN, which is essential for customers using pipelines.

1. Create desktop flow connections with a service principal as the run owner (by default, the run owner is the connection owner).
2. Share the connection with the pipeline SPN.
3. The pipeline exports the solution as the owner of all records of the managed solution.
4. The Pipeline SPN imports the solution into the production environment, maintaining the mapping between connection reference and connection ID.

Allow customers to define lists of credentials and use them in the desktop flow connections. These lists allow customers to define which machine(s) can use these credentials. These lists can be used as login credentials in the desktop flow connections and will benefit from the same credential management feature, such as:

• Encryption and data protection
• Password rotation
• Application lifecycle management

During the public preview, only CyberArk credentials are supported for mapping.

The voice command functionality for Power Automate's Copilot allows you to create and manage desktop flows using natural language spoken through a microphone. You can turn on voice commands by pressing the microphone icon in the bottom right of the Copilot side cart and start dictating automation steps hands-free. To finish, you press the stop icon, and your prompt is automatically submitted to Copilot. This feature reduces manual input, saving time and effort, and enhances Copilot's versatility and user-friendliness, contributing to improved productivity and streamlined operations.

Web client for process mining provides customers with the ability to view and analyze process mining results directly through a web-based interface in Power Automate Process Mining page.

Key functionalities are:

• Replaced default embedded Power BI report with fully interactive Process overview dashboard.
• Process map component offers a full range of KPIs to display.
• Content of Process Overview page is dynamically recalculated based on filters set by the user.
• Power BI report is still generated, but users don't need to wait for its creation to access Process Overview analysis.

Process insights from Power Automate Process Mining can now be consumed in Power BI reports created in Fabric workspace by leveraging the DirectLake semantic model on top of Fabric Lakehouse.

When compared to the semantic model created in Power BI workspaces, the new DirectLake model:

• Is simpler and more streamlined, eliminating unnecessary complexity.
• Uses DirectLake mode, thus requiring smaller SKU even for large process models.
• Less resource intense, as DirectLake mode loads to memory only data required by a particular query.
• Delivers on par performance with import mode semantic model.
• Allows custom enhancing using composite models while still keeping the DirectLake perfomance benefits (subject to limitations of DirectLake composite models).
• Includes the custom metrics and view configuration created in Power Automate Process Mining desktop app.

In the Create process wizard, you can select the desired location for the resulting semantic model:

• Power BI Embedded: Creates a lite version of default report in a provided shared workspace and semantic model is based on aggregated values. This is recommended for small proof-of-concepts and for trial.
• Power BI workspace: Creates a full semantic model (older version with import mode) in your own Power BI workspace. This is recommended for production, small, and mid-sized use cases.
• Fabric workspace (preview): Creates a full semantic model (DirectLake) in your Fabric workspace. This is recommended for any production use case.

This feature is connected to the feature Support for delta-parquet data ingestion and extends the options that you can choose from, with respect to data ingestion. You're now able to ingest data through:

• Dataflows
• Bring your own Azure Datalake Gen2. Data can be in CSV, parquet, and delta-parquet format
• OneLake through Fabric Lakehouse. Files can be in CSV, parquet, and delta-parquet format

The feature is automatically turned on as an option in the Create process wizard. Its goal is to provide ingestion from Microsoft Fabric unified data platform for creating a data twin of an organization.

Today in Power Automate Process Mining, you need some level of technical and product knowledge in order to get to actionable insights. It requires that you create filters, custom metrics, and categorization variables to get to a standardized view of the process. You can use Copilot to get to iterative actions in process mining, process insights, and an introduction to process mining that allow you to start with the tool in an easy way.

Analytical copilot offers:

• Desktop plug-in to create actions on process.
• Ability to summarize findings qualitatively and quantitatively.

Process Intelligence Analytic Service offers analytical functions in the form of an Azure API. When the desktop application connects to Process Intelligence Analytic Service, you have access to the identical functionality as in the full Power Automate Process Mining desktop app.

You can turn this feature on in the Desktop App options.

With Power Automate Process Mining thin client app, there's no need to:

• Download the process model, which takes up your resources.
• Fulfill special hardware requirements when working with large process models.
• Install desktop application updates in case of analytical operation modifications.

Capacity tracking for paid licenses is an important feature for any tenant. The licensing model for Power Automate Process Mining is based on imported/analyses data volumes. Each time a user imports new data into process, the purchased capacity is spend depending on the volume of imported data. This spending is now included in the reporting 'Capacity Utilization' page. The following functions are available:

• The 'Capacity Utilization' page provides information about actual capacity allocation and spending in environment.
• Users can see how much capacity was consumed by their processes and how much by the processes created by other users.
• Information about unlimited amounts, available capacity, or capacity over usage is provided.
• Users can raise a request to the tenant admin to assign more capacity for a given environment.
• Users can get a more detailed report with capacity spending per individual processes.

Power Automate Process Mining connector allows you to bring process mining insights to Power Automate flows. Flow can be triggered based on events fired by business rule evaluation in Power Automate Process Mining. The business rule is evaluated each time new process data are imported. When a business rule is defined in Power Automate Process Mining, users in flow can pick up a new flow trigger, which is initiated by evaluation of selected business rules.

As Power Automate Process Mining is now a cloud connector, the following scenario is available:

• At start, the user defines a business rule in Power Automate Process Mining.
• When a business rule is defined, in cloud flow, you can pick up a new trigger "When business rule is evaluated" for the Power Automate Process Mining connector.
• Trigger returns a list of objects which meets criteria of a selected business rule.
• The list contains objects ID, defined KPI value, and KPI category as defined in business rule.
• You can define automated actions for individual objects.