In-app agents act as a reasoning engine to automate "cognitive" processes like routing, orchestration, and decision-making–removing manual effort from business processes. By directing business users to the next highest value task, these agents enable users to focus their time and energy on the biggest impact things their desired business outcomes, instead of the drudgery of digging through all their data.

Makers configure in-app agents with the types of actions possible for different tables. Then, the agent processes the rows assigned to a specific user to understand the most valuable next action and recommends it.

This feature is available as an early access preview. Sign up for our early access program by filling out this form: https://aka.ms/PowerAppsEAP

As a maker, you can now fully leverage the capabilities of the Plan Designer by creating a plan from your existing solution. The Plan Designer uses AI to extract information from your solution, generating the business problem and user requirements. Once the plan is saved, you can edit the user requirements and generate additional apps, tables, and other components for your solution. To generate a plan, your solution must include at least one app and one associated table, and the apps must be published.

Go to the Solutions menu in Power Apps to create a plan from a solution.

The Solution Explorer is being integrated into the Plan Designer. Makers using Plan Designer will be able to quickly switch back and forth between the high-level Plan view and the detailed-level Objects view showing Solution Explorer. The key benefits are reduced context switching—the maker doesn't have to use multiple tabs to check that the plan and underlying solution components in sync. This is especially helpful as makers manage their solutions for ALM.

How to access the Objects view

1. Create and save a plan.
2. Once the plan is saved, then Objects will be enabled in the left navigation menu.
3. Selecting Objects will bring the user to the embedded Solution Explorer view.

Makers with the roles of System Administrator or System Customizer in their environments will be able to use the Plan Designer.

The Plan Designer will recommend a set of solutions tailored to address your business problem based on your organization's needs. Some of these solutions may already exist in your environment. Instead of creating new apps, you can replace a suggested app with an existing one or simply add an existing app to the set of solutions directly within the Plan Designer.

The Plan Designer feature offers a high-level architecture to address specific business challenges by recommending the creation of various objects, such as apps, flows, and tables. Plan agents are AI assistants that analyze your business problem and generate a plan for it. These plan agents utilize AI to help build your plan. The Solution Agent now suggests existing apps, enabling makers to incorporate these apps into their plans. This enhancement allows makers to utilize the Plan Designer to define the solution architecture for their scenario while reusing pre-existing apps.

• The Solution Agent now provides more comprehensive recommendations by suggesting existing apps.
• Makers can customize their s by adding existing apps and tailoring the architecture to their specific needs.

The Plan designer helps users create a high-level architecture to solve business problems. It suggests creating objects like apps, flows, and tables. Users can replace the suggested apps with their existing ones, allowing them to define the solution architecture for their scenario and reuse pre-existing apps. This customization provides flexibility, while reusing existing apps saves time and resources.

A Copilot Studio Agent created using Plan Designer is designed to streamline workflows, automate repetitive tasks, and answer customer questions. Once the agent is created in Plan Designer, you will be redirected to Microsoft Copilot Studio to review and make any necessary changes before publishing. The agent generated by Plan Designer comes prefilled with the following details:

• Name: The name of the agent.
• Description: A detailed explanation of the agent's purpose.
• Instructions: A set of tasks the agent will perform.
• Knowledge: A list of data sources the agent will use to complete tasks, including all data tables created by the plan.

It is recommended to review the agent's instructions, make any desired changes, and test the agent before publishing.

To build a successful business plan, being able to reuse existing assets, such as existing tables in your solution, is crucial. Today, makers can leverage existing tables in the standalone data workspace experience, but not when they are working with a plan. With this new feature, makers will not only be able to edit new tables as part of their plan's data model but also existing tables. To further boost maker productivity, Copilot will help suggest the right existing table based on the business requirements.

Makers can begin with a description of their business problem and provide additional context like process diagrams, data models, or even screenshots of legacy apps. Copilot will use this input to draft user roles and requirements for the solution. Then, Copilot suggest a multi-table data model according to the user requirements for makers to approve and continue to iterate on. Makers will be able to manually select and edit the existing tables of their plan through the data workspace, as well as chat with Copilot to include the existing tables based on their requirements.

Agent builder for Power Apps empowers makers to create specialized custom autonomous agents that automate repetitive business processes. By understanding the app and business process, Agent builder adds knowledge and triggers to the agent, making it easier than ever to streamline operations with autonomous AI agents.

Makers can effortlessly create agents that automate processes within their existing canvas apps. Agent builder leverages the app's metadata and the desired agent goal to generate a comprehensive step-by-step process. This process is then combined with extracted skills from the app, resulting in a fully equipped copilot that provides detailed instructions and actions.

Are you tired of creating new tables, only to discover that they duplicate existing ones in your environment? Say hello to a smarter way of designing data models using the Copilot-assisted ERD designer. Now, instead of starting from scratch, you can leverage your existing tables to build more robust data models. Here’s how it works: Simply type out your requirements in natural language, and Copilot will craft a multi-table data model that incorporates both new and existing tables. Need to find relevant tables quickly? Copilot’s recommendations have you covered and will locate the right tables in seconds. Once you’ve assembled your tables, switch to the ERD view. There, you can review your new and existing tables side-by-side. While editing existing tables in the ERD view isn't fully supported yet, you can create lookups into the existing tables in a simple drag-and-drop interaction to define meaningful connections between tables.

Creating solutions to address business needs involves more than just developing an app. Solutions are a collection of objects from across the Power Platform that store data, provide end-user apps, and automate workflows. Regardless of the technology used to solve a business problem, the first step to success is defining your solution architecture to support various user roles and their requirements. This ensures you can build a solution that benefits your business. Copilot can now assist with detailing the scenario and requirement definitions, such as user roles, user requirements, data schemas to support data, and creating an overall solution architecture through proposed objects.

• Scenario and requirement definition: Begin with a description of your business problem and provide additional context like process diagrams, data models, or even screenshots of legacy apps. Copilot will use this input to draft user roles and requirements for the solution.
• Data modeling: Once aligned on requirements, Power Apps will propose a data model to support the solution and allow the maker to explore and make updates through a visual entity relationship diagram.
• Solution architecture: Copilot will recommend the types of apps and automation that will fit the requirements. In the initial release of this functionality, the recommendations will include canvas and model-driven apps, as well as cloud flows in Power Automate.

During the early phase of the public preview, these Copilot skills are off by default. To access these features, simply turn on the Try the new Power Apps experience toggle on the Power Apps home page. This will allow you to create and manage your plans.

You can easily find, filter, and sort your data using natural language grid filtering, which eliminates the need for complex, advanced filters by allowing you to describe what you're looking for.

For example, if you ask for cases that are high priority with an overdue follow-up by date, this filters your view to only show those relevant cases. Using natural language saves time as you no longer need to use column filters or advanced filters.

Supported features:

• Filtering of records
• Sorting
• Text search

Please note that this feature is currently in preview. It will be enhanced with each weekly release and can be turned on through the admin setting 'Natural Language Grid and View Search'.

In app designer for model-driven apps, you can use AI to help you generate descriptions for your apps. You will have the option to generate an app description in the Publish dialog or in the App Settings panel. This will create a description that you can review and then save as part of the app. The AI-generated app description is also available to app users during the web player experience by hovering over the app name in the header.

You can now export a canvas app as a binary file with the extension .msapp. You can use Power Platform CLI to convert a .msapp file into source code files. With this feature, you no longer need to use the Power Platform CLI to convert the .msapp files. You can view the source code for the canvas app in an updated, human readable YAML format. Each screen has a separate file.

View the source code of your canvas app directly in Power Apps Studio or use external code editors.

This new capability helps you to:

• Store the files in a code repository.
• Automate generation of a canvas app.
• Improve the generative canvas app models.
• Refactor canvas apps without using the designer.
• Integrate natively with Git for Power Platform.