Microsoft’s September 2026 Power Platform feature update is out, and it marks a meaningful shift in how data and business teams will build apps and automate workflows. The headline: AI agent-driven app creation is now generally available, not just a preview feature you have to wait on.

Here is what changed and why it matters for anyone who builds on the Microsoft stack.

Canvas Apps Now Built Through Agent-Driven Interactions

The canvas authoring agent plugin has moved from preview to general availability. That means makers can now describe what they want an app to do and have the agent handle the scaffolding, rather than wiring up each screen and control by hand.

This is not a small quality-of-life improvement. Canvas apps are notoriously time-consuming to build for anyone without a development background. The authoring agent removes a significant portion of that friction, which opens Power Apps to a wider group of people inside a business who have the domain knowledge but not the maker skills.

For data teams in particular, this changes the conversation around self-service tooling. Instead of waiting for IT or a developer to build a lightweight data entry or reporting interface, the team can describe it and iterate from within Power Apps.

Natural Language to Model-Driven Apps

The update also introduces the model app-builder skill, which does something similar for model-driven apps: describe your business requirements in plain language and the system generates forms, views, and security roles to match.

Model-driven apps are the more structured, database-centric side of Power Apps — the kind used for CRM-style workflows, case management, and anything that needs consistent data structure across teams. They have traditionally required more technical knowledge to build well. The builder skill lowers that bar.

Combined, these two features point toward a near-term future where the primary skill in Power Platform development is knowing what you want to build and being able to describe it clearly, rather than knowing the tooling inside out.

Power Automate Gets Copilot Studio Orchestration

The integration between Power Automate and Copilot Studio is now available, letting automation flows call out to AI agents mid-process. This is the piece that turns straightforward rule-based automations into something more capable.

Think of it this way: a Power Automate flow handling invoice approvals can now hand off to a Copilot Studio agent when it hits an edge case that needs interpretation, then receive a decision back and continue. That kind of conditional AI reasoning inside a business process was hard to wire up cleanly before.

For organisations already running Copilot Studio agents for internal helpdesk or policy queries, this update lets those agents participate directly in automated workflows rather than sitting in a separate chat experience.

Generative Pages Get External Data Sources

Generative pages — AI-generated interfaces within apps — can now connect to Power Platform connectors beyond Dataverse. That includes SharePoint and SQL Server, which covers a large portion of where enterprise data actually lives.

Previously, generative pages were constrained to data stored in Dataverse, which limited their practical use for teams that had not migrated everything into Microsoft’s data layer. This connector support in preview removes that constraint and should accelerate adoption of generative interfaces in real workflows.

What This Means for Business

Microsoft is not just adding features to Power Platform — it is progressively moving the platform toward a model where you describe outcomes and AI handles the construction work.

For data and operations teams, the practical implication is this: the gap between having an idea for an internal tool and actually having that tool is shrinking. The bottleneck used to be maker capacity or developer time. It is now closer to being able to clearly articulate what you need.

A few areas where this will have direct impact:

Internal reporting tools. Teams that previously relied on shared spreadsheets or waited for BI developers to build views can now describe the interface they need and ship it faster.

Process automation with AI decision points. The Copilot Studio integration means automations can handle the straightforward cases at speed and hand off intelligently when judgment is needed.

App governance at scale. As building gets easier, governance becomes more important. The security roles generated by the model app-builder skill are a practical sign that Microsoft is thinking about this — generated apps still need to respect who can see and do what.

The platform is moving in a direction that benefits organisations willing to invest in getting their data and business logic in good shape. The AI features are only as useful as the underlying data they connect to.

---

If your team is running on Microsoft tools and wondering how to get more out of Power Platform, Enterprise DNA’s data training programs cover Power BI, Power Automate, and the broader Microsoft data stack — with courses built for teams that want practical skills, not theory.