Every company runs on its own applications. Applications maintain business rules, permissions, and process knowledge to continue working. We’re making the apps you use every day dramatically more intelligent and productive. With these new updates, Microsoft Power Apps brings AI, Copilot, and agents directly into your apps, extending the intelligence of those apps to the AI surfaces where work happens.
New features include:
- Microsoft 365 Copilot is now generally available in public preview for model-driven and canvas apps, bringing the full intelligence of Copilot into your business process flows.
- new app skills include Data entry, exploration, visualization, summarization now generally available Power Apps. Copilot Chat’s app-based forms and grid experience will be generally available in July 2026, with support for custom UX in preview.
- Agent feed using Power Apps MCP server will be generally available on May 4, 2026, providing a dedicated experience that allows users to monitor agent activity directly within their business apps.
Each of these advancements aims to bring AI to the point where work is actually done. A feature of this release is its interoperability. Agents and Copilot are built directly within the app to increase productivity. Meanwhile, the app’s functionality is built into the agent, helping to ground the AI in a real-life context and make it more trustworthy and trustworthy.
Embed Microsoft 365 Copilot directly within your app
The biggest productivity gains from AI come not from individual tools, but from embedding it within apps where someone is already working on it and where the data model, business rules, and user context all exist.
Now, when a user fills out a form, the app can automatically convert emails and documents into structured fields and perform a verification step before saving. With natural language search, users can simply ask, “Show me the high-priority tickets that are open this week,” and the view is instantly rebuilt. AI-generated summaries extract long activity histories in seconds.
But this is just a starting point. Microsoft 365 Copilot is now generally available for model-driven apps, giving users even more breadth. This means you can ask questions and get answers based on all your business and productivity data through Microsoft Work IQ, not just what you record on your screen. The Copilot experience respects the same security, permissions, and business logic that your app already applies. Administrators enable this at the tenant level. Manufacturers can be configured with just a few clicks. The same app is now much more productive.
Provide app skills to agents
new app skills Data entry, exploration, visualization, summarization, and more are now generally available in Power Apps. But the value of a business application goes beyond its own UI.
App MCP Server allows you to publish these app skills, including structured forms and grid views, in public preview as reusable tools for agents, with additional custom UX skills coming soon. This extends app productivity beyond the app itself to AI surfaces such as Copilot, custom agents, and automations.
This is where the two-way relationship between apps and agents becomes tangible. In the previous section, AI and Copilot were built into the app to speed up the user’s work. Here, the functionality of the app flows externally to the agent and copilot. For example, a recruiting app that has accumulated years of recruiting policies will be able to power agents that access the same records, apply the same rules, and operate under the same control. As organizations digitize more of their processes, agent capabilities will also increase.
Human oversight where it matters most through agent feeds
As agents perform more tasks, the question becomes how to keep humans in the loop at the right times. Agent Feed, a model-driven app that will be generally available in May 2026, directly answers this.
Business users have a dedicated space within the app to see, review, and guide agent activity as it happens, rather than in a separate monitoring tool. Manufacturers control approval thresholds. Low-risk actions complete silently in the background. High-impact actions, such as sending an email, surface as explicit approvals. Side-by-side comparisons, deep links to records, and performance signals make monitoring a reality.
For example, insurance claims teams use agents to extract data from incoming emails and pre-populate case forms. Moderators review and approve anything in the agent feed before it enters the system. Human supervision. Executed by agent. The work is done.
Take AI from experimentation to execution
From embedding intelligence directly into apps, to extending app functionality across AI surfaces like Copilot and custom agents, to enabling seamless human-AI collaboration through agent feeds, all the components are connected. Taken together, this answers the question business and IT leaders are asking: how to implement AI with real impact on core business processes without starting from scratch. This is what makes the transition from experimentation to execution a reality for companies.
See product demos showing how apps, agents, and Microsoft 365 Copilot power business transformation at the Business Applications Update event.
