What’s new in Power Platform: September 2026 feature update by Tiffany Treacy
Saturday September 19^th, 2026 at 7:57 AM

Power Automate Archives - Microsoft Power Platform Blog

Summary Welcome to the Power Platform monthly feature update! We will use this blog to share news in Power Platform from the last month, so you can find a summary of product, community, and learning updates from Power Platform in one easy place. Now, let’s dive into what’s new in Power Platform:

Get started with the latest updates today!

Jump into Power Apps, Power Automate, and Power Pages to try the latest updates, you can use an existing environment or get started for free using the Developer plan.

Join us at PPCC!

See these updates and more come to life at the Power Platform Community Conference 2026 – Join us October 27-29 at the MGM Grand in Las Vegas to discover how organizations are using agents, apps, and automations to drive business outcomes, build new solutions, and scale innovation. Connect with product experts, industry peers, and the community shaping what’s next.

Power Platform

Build hands-on Power Platform and AI expertise with Power Series

Developed by the Power CAT team, Power Series is now available with 20 hands-on labs that help customers, partners, and field teams build practical Power Platform and AI skills at their own pace. Drawing on the team’s experience delivering enterprise customer workshops, Power Series brings that expertise into a reusable, self-service learning experience.

Learners gain hands-on experience choosing the right capabilities, building working solutions, and making design decisions based on their business needs and solution complexity. The outcome: practical skills and greater confidence to apply Power Platform and AI to real projects, from modernizing applications and automating processes to governing solutions.

Learn more

Power Apps

Build polished canvas apps faster with fluent screen templates

Start new canvas app experiences faster with six additional ready-to-use screen templates based on fluent 2 patterns along with recently shipped 3 fluent templates. These templates give makers responsive layouts, editable controls, and practical starting points for common app scenarios. Add one from the new screen experience, connect your data, and customize the design instead of building every screen from scratch.

These templates are a replacement of previous out of box templates built on classic controls.

Additional improvements to modern controls and update from classic to modern controls

Update controls faster. Makers can now update eligible controls in bulk from the control-update surface instead of repeating the process one control at a time. Update all works on the current screen, can enable the required modern-control settings, and preserves the familiar Studio undo and redo experience. Review affected formulas after updating because property names, enum values, and behaviors may change.

More modern controls and quality improvements. The newly released progress bar supports determinate and indeterminate experiences, Power Fx-driven values, semantic colors, theme-aware styling, and accessible progress semantics. The modern avatar and spinner are also available, with improved interaction and upgrade behavior. Continued quality work makes date picker interactions and reset behavior more reliable and preserves formulas that reference checkbox, rating, and toggle.

Learn more

Agentic apps

Generally available: refreshed model-driven apps UI, plus display density in public preview

Model apps UI modernization: Header and navigation refresh reaches general availability, with Display density arriving in preview

The next wave of UI modernization for model-driven apps arrives with version 2609.1. The header and navigation refresh feature reaches general availability, and the new display density feature enters preview. Together they deliver a cleaner, more efficient layout across the app shell, forms, and views, increasing your working area, reducing the time you spend navigating pages, and aligning the experience with modern Microsoft 365 design patterns.

Header and navigation refresh brings a modern app header with simplified layout and improved spacing, a streamlined sitemap that’s easier to scan, and noticeably more working space on forms. The command bar is now the only element fixed at the top of the page, with the summary area and form header scrolling alongside the rest of the form. Once the form header scrolls out of view, a condensed sticky header attached to the bottom of the command bar is shown. At GA, the experience remains opt-in so makers control adoption. Apps that already have the feature enabled get these enhancements automatically.

Display density (preview) answers a longstanding customer request to fit more content into model-driven app pages. Three levels are available — comfortable (the default), cozy, and compact — so users can tighten the interface to match how they work. An app setting lets customers set the default level for their users or turn the feature off entirely, and users can adjust their own density through personal settings. Display density requires the header and navigation refresh feature and is enabled by default once that feature is on.

Learn more

AI powered development

Generally available: canvas authoring agent plugin for AI-assisted app building

The canvas authoring agent plugin is now generally available, giving makers and developers a new way to work with Power Apps canvas apps using agent-assisted development workflows. With the authoring agent plugin, makers can more easily create, inspect, and update canvas apps through agent-driven interactions with their favorite coding agent, helping accelerate common app-building tasks while keeping makers in control. Learn more about the agent here.

This update is part of our continued investment in making canvas app creation faster, easier, and more approachable. By bringing agentic authoring and MCP support to canvas apps, makers can use modern AI development patterns to move from idea to working app more quickly, iterate on app structure and controls, and streamline parts of the authoring process that previously required more manual effort.

Learn more

Generally available: vibe code entire model-driven apps with the app-builder skill

Vibe code entire model-driven apps with the new app-builder skill (preview)

Generative pages gave makers an AI-assisted way to build individual pages in model-driven apps. The new model app-builder skill extends that approach to the whole application, and it’s now generally available. Used with an AI code generation tool such as GitHub Copilot CLI or Claude Code, it builds and edits model-driven apps from natural-language requirements.

You start by describing the business process or app you need. The skill turns your requirements into an application plan that you review before it makes any changes. It starts with user personas and the jobs they need to accomplish, and then extends to cover a growing list of model-driven app artifacts including:

Tables, columns, relationships, and sample data
Forms, views, and charts
Generative pages for experiences that go beyond standard forms and views
A sitemap with custom icons for each table
JavaScript validation rules on forms
Security roles based on the planned personas and data access needs
Business process flows (recently added)
Business rules (recently added)

No app artifacts are created until you approve the build plan. The skill also works on editing your existing apps. Everything it produces is a standard Power Apps and Dataverse artifact, so you can keep iterating with the skill or switch to the Power Apps studio designers whenever you prefer.

Learn more

Embed a generative page directly inside a model-driven app form

Generative pages are no longer limited to standalone pages in your app’s navigation. You can now add a generative page to a model-driven app form so that it appears within a section or tab, letting an AI-generated experience sit right alongside the standard fields on a record.

Build the generative page to accept the recordId input parameter, and when a user opens a record, the form automatically passes the current record ID to the generative page for you. The page knows which record it’s on from the moment it loads. This unlocks the mixed layouts makers have been asking for, keeping the out-of-box form for structured data entry and dropping in a purpose-built generative page for the part of the record that needs a richer, more tailored view.

Learn more

Public preview: generative pages can now use data from Power Platform connectors

To date, generative pages have been built on Dataverse tables, but now they can reach further. Connector support (preview) lets a generative page use data outside of Dataverse through the Power Platform connector ecosystem. If the data your scenario needs already lives in SharePoint, SQL, or any other connected service, you can build a page against it directly.

Connector support works in both authoring experiences by targeting existing connection references in the environment. In the generative page designer, configured connectors are available through add data > connectors, where you select the connection reference and choose the data the page should use. With AI code generation tools, describe the data you want, and the agent walks you through selecting the right connector and adding the binding so it deploys with the page.

Learn more

Power Automate

Quick start cards make it easier to begin automating from the Power Automate home page

New quick start cards on the Power Automate home page provide clear, actionable entry points for common automation scenarios. Users can move directly into creating a flow without first navigating menus or searching through the template gallery. By bringing relevant starting points to the home page, this experience helps new users discover what they can automate while giving experienced users a faster path from idea to flow.

Server-side search helps users find flows faster

Searching in My Flows is now faster and more reliable with server-side search. Instead of filtering only the flows already loaded in the browser, Power Automate sends the search request to the service and returns matching flows from the complete list of available flows in the current environment. This improvement is especially helpful for users who manage large numbers of flows. It provides more complete results while reducing the time and effort required to locate a specific flow.

Learning updates

Training paths and labs

Updated training

Power Apps maker

New

Updated

Power Apps user

New

Updated

Power Automate

New

Updated

Power Pages

New

Updated

Power Platform administration

New

Updated

Power Platform developer

New

Updated

AI Builder

Updated

Power Platform connectors

Updated

The post What’s new in Power Platform: September 2026 feature update appeared first on Microsoft Power Platform Blog.

Read the whole story

alvinashcraft

11 hours ago

Pennsylvania, USA

BONUS The Hidden Dangers of AI at Work With Ari-Pekka Skarp by Vasco Duarte
Saturday September 19^th, 2026 at 7:56 AM

Scrum Master Toolbox Podcast: Agile storytelling from the trenches

BONUS: The Hidden Dangers of AI at Work With Ari-Pekka Skarp

AI is usually sold as a productivity tool, but Ari-Pekka Skarp argues that the real story is what it does to the conversations, skills, and purpose that hold teams together. In this BONUS episode, Ari-Pekka explores why organizations are rushing to use AI "as efficiently as possible" without defining what efficiency means, and what that rush is quietly costing us.

Organizations Are Conversations

"The organizations are actually conversations, conversational patterns between people."

Ari-Pekka's path from software engineering in 1999 to psychology, psychotherapy, and change leadership was driven by one thread: how the mind works, both individually and socially. Meeting Ralph Stacey, Esko Kilpi, and Douglas Griffin at Nokia changed how he saw organizations. Instead of a machine made of parts, an organization is a living pattern of conversations — people responding to each other's gestures, again and again. George Mead added the idea that the human mind itself is not individual but relational. This matters for AI because a large language model is a new kind of player in those conversations, not just a tool that moves data between them.

The Efficiency Fetish

"It's like how much people are pressing the acceleration pedal in the car. It doesn't tell anything where the car is going."

Many organizations are trying to use AI "as efficiently as possible," but Ari-Pekka points out that few have defined what efficiency means. What he sees instead is measurement of AI usage itself — how many people are prompting, how many tokens are flowing. He calls this tokenmaxxing. The car metaphor is the key: pressing the accelerator harder says nothing about direction, and going fast in the wrong direction is more costly than going slow. Efficiency only has meaning against a purpose, and purpose is itself a conversational achievement — something a team has to talk its way into.

De-Skilling Is the Hidden Cost

"If there's nobody in the room who could review what AI has produced and say whether it's correct or not, it's not an AI strategy. It's a liability."

The risk Ari-Pekka worries about most is de-skilling. When we offload cognitive work to AI, we lose the friction that builds learning. There is neurological evidence that people who rely heavily on AI do not develop the same brain structures as those who work through challenges manually. Some skills are fine to lose — nobody needs machine code anymore — but the ability to review and judge AI output is critical, and it is exactly what erodes when we skip the slow work. The result is a double bind: senior experts burn out under the review burden of fast-produced AI output, while juniors never get the time to build the expertise they would need to review it.

We Need Speed Limits for AI

"We can't optimize individual going as fast as possible... we need a collective... boundaries for individuals."

Ari-Pekka reaches for a historical analogy. Our biological rate of processing information is roughly ten bits per second, and it is not going to change. When we only had horses, we did not need speed limits. When we built cars that could go 200 kilometers per hour, we had to invent rules and boundaries to protect the system. AI is the same: we have reached a threshold where optimizing individual output — more code, more stories, more messages — can damage the whole organization. The control mechanism Ari-Pekka proposes is cognitive friction, deliberately added back into the system so that speed serves the system rather than breaking it.

The Tokenization of Work

"It's very easy to lose the purpose where you are going if you are only doing fragments of work."

Ari-Pekka's article The Tokenization of Work describes what happens when the unit of work is no longer a job, a profession, or even a task. Digital tools make it easy to fragment work into tiny pieces and spread them across AI agents, and in the process the boundaries that gave work its meaning vanish. Purpose is what protects us from burnout: with a clear purpose, people can do very demanding work without burning out, because the work feeds them. Without purpose, exhaustion arrives fast. For Scrum Masters, this means grounding the work in why it matters is more important now, not less.

AI Is an Echo Chamber, Not a Mirror

"The AI is more kind of an echo chamber in a sense that it doesn't push back so much."

Ari-Pekka compares AI to George Mead's "generalized other" — the internalized sense of how others see us. AI can play that role, but with a dangerous twist: it is programmed to be agreeable, so it behaves more like an echo chamber than a mirror. Real people push back, point out mistakes, and keep disagreeing. That friction is where learning, competence, and self-awareness grow. Ari-Pekka's practical move is to prompt AI for three different and conflicting perspectives rather than one, using it to go wider rather than only faster. It is not a perfect fix — the model still tries to merge them into one — but it is better than a single agreeable answer.

A Three-Second Pause

"Take a three-second pause... and just ask yourself, what are you doing?"

Ari-Pekka leaves listeners with a small challenge. A few times a day, when you are about to prompt an AI, pause for three seconds and ask what you are actually doing: are you seeking information, or seeking confirmation? Then consider whether it would be better to call a person and have a real conversation. It is a tiny practice, but it points at the whole episode's message: AI is not neutral infrastructure. It changes the conversations, the skills, and the purpose of work, and the people who notice that — Scrum Masters and Agile coaches especially — are the ones who can keep it from quietly reshaping their teams.

About Ari-Pekka Skarp

Ari-Pekka Skarp is a psychologist, psychotherapist, Lead Change Coach, organizational psychologist, and author. He wrote Mindfulness, mielenselkeys ja myötätunto, hosts Mielen laboratorio, and researches nondualism. His work connects psychology, complexity, Agile, and AI at work.

You can link with Ari-Pekka Skarp on LinkedIn. You can read Ari-Pekka's Finnish writing at tietoisuustaidot.com and his English blog at Fractal Sauna. You can also find Ari-Pekka's previous Scrum Master Toolbox Podcast episodes on his guest page.

Download audio: https://traffic.libsyn.com/secure/scrummastertoolbox/20260919_Ari-Pekka_Skarp_BONUS.mp3?dest-id=246429

Read the whole story

alvinashcraft

11 hours ago

Pennsylvania, USA

Human vs. Agent Reliability Over Long Horizons – How Can We Do What They Can’t? by codemanship
Saturday September 19^th, 2026 at 7:56 AM

Codemanship's Blog

Multiple studies find that – without exception – the reliability of autonomous agent workflows decays geometrically over many steps. There are no examples of complex, reliable software created autonomously by agents – or anything even close.

On a per step basis, humans are – of course – equally unreliable. And yet, somehow, there are many examples of complex, reliable software created by humans.

Despite being just as fallible, humans are able to stabilise reliability in a way LLM-based agents can’t – despite all our best attempts to make them “self-correcting” or “self-healing”.

I think the missing piece of the jigsaw lies in my simplified model of the reliability of a step in development workflows:

R = 1 – (1 – C)(1 – P)

Where C is the probability of it being correct, and P is the probability of any errors being caught before they propagate and compound – before the wheels start to wobble.

I think what we’re looking for is in 1 – P. If P represents the distribution of errors that get caught, then 1 – P represents the errors that fall outside that distribution. These are the errors that weren’t anticipated – the tests the agent didn’t write, the rules the linter didn’t check, the things the model wasn’t trained on.

The state of the art in self-correcting systems can only correct problems if they recognise them as problems. They can only self-correct if the problem is in P.

It’s the gap between P and 1 that causes the instability in autonomous workflows, and there is always a gap. P = 1 is infinitely difficult to achieve on any non-trivial problem.

Humans must have some capability that compensates – some ability to recognise and then adapt to the new and the unfamiliar, and with sparse data to go on.

Cognitive science calls it “learning”.

Read the whole story

alvinashcraft

11 hours ago

Pennsylvania, USA

Job listings for week ending 9/18 by Leon
Saturday September 19^th, 2026 at 7:55 AM

Leon Adato

Job postings that came across my desk, slack, email, discord, etc this week.

The post Job listings for week ending 9/18 appeared first on Leon Adato.

Read the whole story

alvinashcraft

11 hours ago

Pennsylvania, USA

Call For Papers Listings for 9/18 by Leon
Saturday September 19^th, 2026 at 7:55 AM

Leon Adato

A collection of upcoming CFPs (call for papers) from across the internet and around the world.

The post Call For Papers Listings for 9/18 appeared first on Leon Adato.

Read the whole story

alvinashcraft

11 hours ago

Pennsylvania, USA

Gestalt Principle: Common Region (Grouping by a shared boundary) by hello@ux360.design (Shirish Shikhrakar)
Saturday September 19^th, 2026 at 7:55 AM

ux360.design

The law of common region: a shared boundary groups whatever it encloses. Why it beats proximity, what counts as a boundary, and what every boundary costs.

Read the whole story

alvinashcraft

11 hours ago

Pennsylvania, USA

What’s new in Power Platform: September 2026 feature update by Tiffany Treacy Saturday September 19th, 2026 at 7:57 AM

Get started with the latest updates today!

Join us at PPCC!

Power Platform

Build hands-on Power Platform and AI expertise with Power Series

Power Apps

Build polished canvas apps faster with fluent screen templates

Additional improvements to modern controls and update from classic to modern controls

Agentic apps

Generally available: refreshed model-driven apps UI, plus display density in public preview

AI powered development

Generally available: canvas authoring agent plugin for AI-assisted app building

Generally available: vibe code entire model-driven apps with the app-builder skill

Embed a generative page directly inside a model-driven app form

Public preview: generative pages can now use data from Power Platform connectors

Power Automate

Quick start cards make it easier to begin automating from the Power Automate home page

Server-side search helps users find flows faster

Learning updates

Training paths and labs

Power Apps maker

Power Apps user

Power Automate

Power Pages

Power Platform administration

Power Platform developer

AI Builder

Power Platform connectors

BONUS The Hidden Dangers of AI at Work With Ari-Pekka Skarp by Vasco Duarte Saturday September 19th, 2026 at 7:56 AM

BONUS: The Hidden Dangers of AI at Work With Ari-Pekka Skarp

Organizations Are Conversations

The Efficiency Fetish

De-Skilling Is the Hidden Cost

We Need Speed Limits for AI

The Tokenization of Work

AI Is an Echo Chamber, Not a Mirror

A Three-Second Pause

About Ari-Pekka Skarp

Human vs. Agent Reliability Over Long Horizons – How Can We Do What They Can’t? by codemanship Saturday September 19th, 2026 at 7:56 AM

Job listings for week ending 9/18 by Leon Saturday September 19th, 2026 at 7:55 AM

Call For Papers Listings for 9/18 by Leon Saturday September 19th, 2026 at 7:55 AM

Gestalt Principle: Common Region (Grouping by a shared boundary) by hello@ux360.design (Shirish Shikhrakar) Saturday September 19th, 2026 at 7:55 AM

What’s new in Power Platform: September 2026 feature update by Tiffany Treacy
Saturday September 19^th, 2026 at 7:57 AM

BONUS The Hidden Dangers of AI at Work With Ari-Pekka Skarp by Vasco Duarte
Saturday September 19^th, 2026 at 7:56 AM

Human vs. Agent Reliability Over Long Horizons – How Can We Do What They Can’t? by codemanship
Saturday September 19^th, 2026 at 7:56 AM

Job listings for week ending 9/18 by Leon
Saturday September 19^th, 2026 at 7:55 AM

Call For Papers Listings for 9/18 by Leon
Saturday September 19^th, 2026 at 7:55 AM

Gestalt Principle: Common Region (Grouping by a shared boundary) by hello@ux360.design (Shirish Shikhrakar)
Saturday September 19^th, 2026 at 7:55 AM