We've partnered with three blogging platforms built on AT Protocol to make it even easier to get writing. Sign up for any annual plan on Offprint, Leaflet, or pckt.blog and you'll get 25% off, on us. That covers a whole summer's worth of blogging!

Since we launched our Standard.site integration into the Bluesky app, we've been thrilled to see so many writers adopting this community-built format for longform writing. And people are building new apps, like Standard Reader, to take it all in.

If you're ready to take your posting to the next level, start that food blog, give your hottest takes a bit more room, or just have more fun online, now's a great (and financially prudent) time to do it.

Get 25% off your first year when you sign up for an annual plan from:

• Offprint
• Leaflet
• pckt.blog

Hello Folks!

Picture this. A regulator hands you a one-pager that says, in essence, “this data does not leave the building.” Or your link to Azure decides to take a nap during a critical batch run. Or you are standing up infrastructure in a remote site where connectivity is a coin flip on a good day. For a long time, our answer to that conversation was a stack of Azure Stack boxes plus a lot of wishful thinking. That story has changed, and it has changed quite a bit.

At Microsoft Azure Infra Summit 2026, Thomas Maurer (Global Black Belt for Sovereign Cloud) walked us through what is now called the Microsoft Sovereign Private Cloud, with Azure Local as its foundation. In this post, I want to unpack the session for the ITPros in the room, the folks who have to actually run this stuff on Monday morning. Let us dig in.

📺 Watch the session:

Why IT Pros Should Care

Sovereignty is no longer a niche conversation. Thomas was very clear that there is no one-size-fits-all answer, and that is exactly why this matters to us as operators. The drivers landing on our desks now include:

• Regulatory requirements that demand data residency or full operator isolation.
• Sovereign AI workloads where the model and the data both need to stay in-country.
• Disconnected and air-gapped sites by design (think defense, manufacturing floors, retail backrooms, ships, mines).
• Business continuity, meaning a workable Plan B if the public cloud is unreachable for hours or days.
• Latency-sensitive workloads where the round trip to a region is just too slow.

If you build or operate infrastructure that touches any of those bullets, Azure Local is now a first-class option, not a sidecar. And it gets you a cloud-consistent control plane on top of hardware you can put your hands on.

What is Azure Local and the Sovereign Private Cloud

Let us level-set on the stack, from the metal up.

• Hardware. Validated and certified through the Azure Local solution catalog, delivered by the OEMs you already buy from. Form factors range from single-node edge boxes up to multi-rack deployments. There is a Premier tier with extra testing, packaged firmware and driver updates, and AI-ready GPU configurations done with NVIDIA.
• Software-defined data center. Compute, storage, networking, and high availability. As of April 2026, supported SAN storage is GA alongside the existing hyperconverged storage spaces direct model. That gets you up to 64 nodes in disaggregated mode and 16 nodes in hyperconverged mode per instance.
• Workload plane. Linux and Windows VMs, custom images, your own Kubernetes distribution, or AKS enabled by Arc with the same management experience you have in Azure today.
• Arc-enabled control plane. This is where Azure Local stops being “another on-prem stack” and starts feeling like Azure. Defender, Azure Monitor, Azure Update Manager, Policy, RBAC, Resource Manager, all of it surfaces against your on-prem instance.
• Disconnected operations. Microsoft packaged a subset of the control plane (portal, Resource Manager, key management services) into an appliance you deploy on-premises. Connect your Azure Local infrastructure to the local appliance instead of public Azure, and you have a fully air-gapped deployment with a familiar API surface.

On top of that base, the Sovereign Private Cloud bundles workloads you can run locally: Foundry Local for AI inferencing, Microsoft 365 Local (Exchange Server, SharePoint Server, Skype for Business Server) for productivity fallback, Azure Virtual Desktop on Azure Local for VDI, and GitHub Enterprise Local (in private preview at the time of the session) for source and CI/CD.

How it works in production

In the demo, Thomas drove the whole show from the Azure Arc Center in the Azure portal. A few things stood out for me as someone who has spent too many late nights patching clusters.

• One pane, many sites. The overview page rolls up every Azure Local instance you own. Thomas mentioned customers running thousands of these things, and the Azure Local Lens workbook in Azure Monitor is built to manage at that scale.
• Resources feel like Azure resources. An instance, a node, a VM, an AKS cluster, they all live inside Azure Resource Manager. RBAC, activity logs, tags, ARM templates, everything you expect.
• Update is a single button. The Solution Builder Extension packages OS, management software, drivers, and firmware into one validated update. You hit “update,” it orchestrates live migrations node by node, and it blocks the operation if something is not ready. No more cherry-picking driver bundles at 2 AM.
• Security defaults are real. BitLocker on OS and data volumes, SMB signing, App Control on the hypervisor hosts, drift detection that flags configuration changes back to the portal.
• Resiliency is layered. Storage spaces direct two-way or three-way mirroring, rack-aware clustering, live migration for maintenance, and Azure Site Recovery for site-to-cloud replication (currently preview). Site-to-site ASR between two Azure Local instances is in development. Veeam, Rubrik, and Commvault all integrate for backup.

In short, the boring operational moments are the ones that benefit the most. Patching, monitoring, identity, alerting, they collapse into the tools you already use in Azure.

When to use it and real-world scenarios

This is not a “rip everything out of Azure” pitch. Thomas was very honest. Azure is still the right home for the vast majority of workloads. Azure Local earns its keep in a few specific places.

• Regulated or sovereign workloads. Government, defense, financial services, healthcare where the law or the contract says the data stays put.
• Disconnected or air-gapped sites. Field operations, classified networks, ships, mines, remote infrastructure where reliable connectivity is not in scope.
• Business continuity for productivity. Microsoft 365 Local as a fallback for Exchange and SharePoint if the cloud service is unreachable. From the session Q&A, M365 Local is GA, and it is the Exchange / SharePoint / Skype for Business trio. Entra ID and Intune are not in scope of the local bundle.
• Edge and latency-bound workloads. Manufacturing line control, retail in-store inference, healthcare imaging, anywhere a 30-millisecond round trip is a problem.
• Sovereign AI. Foundry Local on Azure Local lets you serve models on local GPUs without round-tripping to the cloud. Models stay local, data stays local, inference stays fast.
• Bi-directional workload mobility. With Sovereign Private Landing Zones, you design once and keep workloads portable between Azure and Azure Local based on a service-compatible subset.

Getting Started

If you are picking this up cold, here is a sensible on-ramp:

• Start with the official docs on Sovereign Private Cloud and Azure Local. Read them with your architect hat on, not just your operator hat. Design matters here.
• Browse the Azure Local solution catalog and filter by Premier solutions and by your target scenario (disconnected operations, M365 Local, AI workloads, GPU support). The hardware shape drives a lot of downstream decisions.
• Talk to your OEM about a validated node, and talk to your Microsoft account team or a sovereign partner. The partner ecosystem in this space is mature, and they will save you weeks.
• Stand up a small connected instance first to learn the Arc Center experience, the update flow, and Azure Monitor integration. Even a one-node or two-node lab is enough to internalize the model.
• For disconnected, size for the extra capacity the control plane appliance needs, plan your local identity (Active Directory with AD FS) and your local monitoring integration up front.
• If you live in Azure today and need workload portability, look at Sovereign Private Landing Zones so you do not paint yourself into a corner with services that have no on-prem equivalent.

Resources

• What is Sovereign Private Cloud? on Microsoft Learn
• Azure Local documentation
• Disconnected operations for Azure Local
• Azure Arc product page
• Azure Site Recovery product page
• Foundry Local documentation on Microsoft Learn
• Foundry Local on GitHub
• Sovereign Landing Zones on GitHub

Watch the rest of the Summit

This was just one of the sessions at the Microsoft Azure Infra Summit 2026. If you want more peer-to-peer technical content from the Azure infrastructure community, grab a coffee and queue up the full playlist here: https://aka.ms/MAIS/2026-Playlist

There is plenty of good stuff covering Bicep, AKS networking, storage, IaC, and more.

If you spin up an Azure Local instance after watching the session, or if you are already running one in anger, drop a comment and let me know how it goes. What works, what hurts, what you wish was better. That is how we all level up.

Cheers!

Pierre Roman

Your agent works beautifully on your laptop. It reasons, calls tools, and nails the demo. Then comes the part nobody enjoys: making it enterprise-ready. Identity. Observability. Governance. Messaging. Secure access to Microsoft 365 data. Until now, that last mile meant reading through Microsoft Learn articles, running CLI commands in exactly the right order, hand-editing manifests, and hoping every prerequisite was in place before the agent reached production.

Organizations are moving fast — from experimenting with AI agents to putting them to work across real business processes. Developers are building with LangChain, OpenAI, Semantic Kernel, Azure AI Foundry, Claude SDK, Google ADK, and more. But great orchestration logic isn’t enough. Agents also need identity, observability, governance, messaging, testing, and secure access to Microsoft 365 data.

Agent 365 Skills collapse that workflow into a guided, natural-language experience inside the coding assistant developers already use. Describe what you need — “add observability,” “wire up WorkIQ Mail and Calendar,” “test this agent locally” — and the skill detects your project, asks only the questions it has to, applies the right changes, and validates the work before the session ends.

The last mile is the hard mile

Before Agent 365 Skills, onboarding an agent often meant a long, fragile chain of manual steps: installing the Agent 365 CLI, validating Azure prerequisites, registering a blueprint, configuring Entra artifacts, wiring OpenTelemetry exporters, packaging manifests, connecting MCP servers, and testing locally. Miss one step, and things break — often late, in production.

From a multi-hour, order-sensitive gauntlet to one guided sentence.

Agent 365 Skills replace that disconnected workflow with a guided lifecycle. Developers stay in Claude Code, GitHub Copilot CLI, or VS Code agent mode and ask for the outcome they want. The skill handles the implementation details — while preserving the technical controls enterprise teams depend on.

What changes for developers:

• Start with one sentence — “add WorkIQ tools” or “integrate observability” — instead of reading many Microsoft Learn pages.
• Auto-instrument in minutes instead of manually wiring OpenTelemetry exporters, token resolvers, and baggage scopes.
• Let the skill apply the required manifest and configuration updates instead of hand-editing them.
• Catch issues before shipping with built-in validators instead of hoping nothing was missed.

Watch it work

Say a developer wants telemetry. They type a single line:

> add observability to this agent

From there, the skill takes over. It detects the framework — say, a .NET Semantic Kernel project — asks only the questions that matter (which authentication mode: OBO, Agentic-User, or S2S), wires OpenTelemetry and the Agent 365 tracing exporter, and confirms the project still builds. No manifest spelunking. No half-configured exporters.

Every skill follows the same five-beat loop: describe, detect, ask, apply, validate.

Other requests work the same way:

• “Add messaging and notifications to this agent.”
• “Add observability to this agent.”
• “Wire up WorkIQ Mail and Calendar.”
• “Test this agent locally.”

Six skills, one lifecycle

Agent 365 Skills include six skills that map to the lifecycle steps developers hit when preparing an agent for Microsoft Agent 365. Run only the ones your agent needs — or use the make skills to execute all the required steps.

Built for the stack you already have

Agent development isn’t one language or one framework, and Agent 365 Skills are designed for that reality. They support three major ecosystems and many of the most popular agent frameworks.

• .NET: Agent Framework, Semantic Kernel, and Azure AI Foundry.
• Node.js: LangChain, OpenAI Agents SDK, Claude SDK, Semantic Kernel, and Google ADK.
• Python: Agent Framework, LangChain, OpenAI, Claude, Semantic Kernel, and Google ADK.

The project stack is auto-detected, so you don’t start by writing configuration files or mapping your project by hand. The skill determines the path and guides the next step.

Enterprise-ready by design

Speed matters — but so does not breaking things. Agent 365 Skills are built to move fast without giving up the safety and validation teams expect.

• Additive and idempotent: Skills don’t delete or restructure existing code, and re-running a skill is safe by design.
• State-aware: A workspace detection cache tracks what’s already wired, so completed steps are skipped, not duplicated.
• Guard-railed: Pre-tool hooks block modifications outside the project, and stop-hook validators verify the build passes before the session ends.

Together, these controls reduce setup risk while giving developers a repeatable path to register agents, add observability, connect Microsoft 365 data, and test locally.

What your agent gains

One onboarding flow connects your agent to identity, observability, data, messaging, and testing.

For customers, Agent 365 Skills shorten the path from prototype to enterprise-ready agent. Developers spend less time stitching setup steps together and more time building agents that solve business problems. IT and security teams get more consistent registration, observability, and governance patterns across the agent estate.

The result is a faster, safer onboarding experience: agents registered with Agent 365, instrumented for visibility in Defender, Purview, and the Microsoft 365 admin center, connected to Microsoft 365 data through Work IQ MCP servers, and tested locally before deployment.

Get started

Install the skills with:

gh skill add microsoft/agent365-skills

• Explore the GitHub repository.
• Read the Microsoft Agent 365 developer docs for deeper developer guidance.

Healthcare and life sciences teams are asked to do the impossible every day: deliver better outcomes, move faster, and stretch every dollar — all while navigating some of the most regulated, documentation-heavy workflows in any industry. The promise of AI has never been about replacing the experts who do this work. It’s about giving them their time back.

That promise is entering a new phase. We’re moving from AI that answers to AI that acts — agentic AI that can carry out multi-step work across your applications, your documents, and the web, with you in control. Nowhere is the opportunity more concrete than in the daily operational grind of healthcare and life sciences.

From answering to doing

Most knowledge work in HLS isn’t blocked by a lack of information — it’s blocked by the effort of pulling that information together and turning it into action. Gathering the right documents. Summarizing a thread. Drafting the first version. Updating five systems with the same three facts.

Microsoft Scout is designed for exactly this layer of work. Think of it as an agentic AI teammate on your desktop — one that can read and organize files, search across your email, calendar, and Teams, browse the web, and complete genuinely multi-step tasks on your behalf. Crucially, it can run on a schedule, so routine work happens before you sit down, and it keeps a human in the loop for anything that matters.

Real-world ways HLS teams can do more with less

• Provider operations: Assemble the documentation needed for a referral or prior-authorization request, summarize the relevant history, and draft the submission — turning a 30-minute scramble into a two-minute review.
• Clinical research coordination: Pull together study start-up documents, track outstanding site communications, and draft consistent follow-ups, so coordinators spend their time on sites and patients rather than inboxes.
• Medical affairs and field medical: Prepare for an engagement by gathering the latest publications and prior interactions into a single brief, then capture a structured summary afterward — every meeting, consistently.
• Commercial and market access: Stand up an account briefing or a competitive news roundup on a recurring schedule, so the team starts every week informed instead of researching from scratch.

The thread running through all of these is the same: reclaim capacity, increase consistency, and accelerate speed to market — doing more with the people and budget you already have.

Built for the trust HLS demands

In this industry, “helpful” is not enough; it has to be trustworthy. Agentic AI for healthcare and life sciences has to respect enterprise security and data boundaries, keep sensitive information under your control, and keep a person in command of consequential decisions. The goal is to automate the busywork around expert judgment — never to automate the judgment itself.

A family of AI that works the way you do

Scout is part of a broader shift in how Microsoft is bringing AI to work. Where Microsoft 365 Copilot brings AI into the flow of the apps you already use, and Microsoft Cowork reimagines how teams collaborate with AI, Scout focuses on agentic action at the desktop — automating end-to-end tasks and recurring workflows. Together they point to the same future: more of your day spent on the work only you can do.

Start small, compound the gains

You don’t need a transformation program to begin. Pick one repetitive, high-friction workflow — the weekly roundup, the recurring briefing, the documentation prep that nobody enjoys — and let agentic AI take the first pass. The time you reclaim funds the next idea.

We’ll be sharing practical, healthcare- and life-sciences-specific playbooks here every week. Subscribe to the blog, and tell us in the comments: what’s the one recurring task you’d hand to an AI teammate first?

Learn more about Microsoft Scout:

• Introducing Microsoft Scout: Your always-on personal agent | Microsoft 365 Blog 
• Microsoft Scout (Frontier) overview | Microsoft Learn 

SharePoint is getting a refreshed look and feel designed to make the product feel simpler, clearer, and more focused on your content. We’ve recently introduced a new SharePoint experience, organized around helping you discover knowledge, publish content, and build solutions. Along with that new experience, we’re introducing visual updates across key product surfaces that reduce visual noise, improve readability, and bring more consistency to the interface while preserving the branding investments your organization has already made. We call this work the SharePoint visual refresh: a thoughtful update to the product’s visual language that helps SharePoint feel more modern, approachable, and easier to use.

Background

Over time, SharePoint’s capabilities have continued to grow, which has given us an opportunity to update the look and feel as well. These updates not only make the SharePoint UI fresh, but it ensures that we are consistent across M365.

The SharePoint visual refresh, rolling out to general availability now, is part of a broader collaborative effort across Microsoft 365 to modernize the suite and improve usability with this bold, new visual design, while also responding to customer feedback about visual clutter and the need for clearer focus on content. These updates are grounded in core design principles that prioritize usability, coherence, and a sense of delight in everyday interactions.  This bold, new design is part of a larger update across many apps (such as Microsoft 365 Copilot) and platforms that family together to create a cohesive, modern look and feel.

Goals
The Visual Refresh is guided by several key goals:

• Deliver a more contemporary, polished look aligned with the broader Microsoft 365 experience to existing and new SharePoint experiences
• Improve consistency, usability, and accessibility across key SharePoint surfaces, including Sites, Pages, and Document Libraries and the new Discover, Publish, Build destinations.
• Reduce visual clutter to improve focus on what matters most: your content
• Create a design system that scales, from simple team sites to complex enterprise scenarios 

Consistency Across Experiences

Once adopted, the visual refresh will strengthen alignment between SharePoint, OneDrive, Teams and the Microsoft 365 shell, reducing visual fragmentation across navigation, pages, and components.

By aligning how our UI looks and our theming approaches, we aim to create a more predictable experience for users moving between tools and surfaces, helping avoid visual dissonance and reducing cognitive load across day‑to‑day workflows. 

What’s being updated

1. Canvas elevation: putting content in the spotlight
   One of the most noticeable changes is the elevation of the SharePoint site canvas. By adding a subtle shadow, refined spacing, and a light gray shade to backgrounds, the refreshed design helps your content stand out, making it easier to scan, read, and interact with pages.
   

   Importantly, this is not achieved by changing your content. Instead, the improvements come from thoughtful adjustments to padding, layout spacing, and visual framing, ensuring that the emphasis remains on the information you create and manage. This does not effect existing layouts or how sites reflow.

   
   Page view before / after Visual Refresh

2. Neutral theming for app surfaces
   We’re also evolving how themes are applied within SharePoint to improve consistency, accessibility, and clarity across experiences by separating the core SharePoint user interface from customer content. Updated neutral theming of app surfaces provides a more stable visual foundation and establishes a unified set of navigational UI, where the customer brand remains in an anchor position and is in harmony with the new neutral navigation. This approach allows customer branding to be expressed more clearly without competing with structural UI elements, clarifies the distinction between customer branding and the SharePoint app, reduces visual competition, and improves focus on primary content.
   Previously, surfaces were a mix of SharePoint and Customer theming; simplifying how surfaces are themed improves clarity and reduces visual noise, improving usability.

3. Styling updates
   Additional updates to typography, spacing, and corner rounding introduce a more cohesive and contemporary visual language across SharePoint surfaces. Updated typography and spacing enhance readability and create more consistent rhythm across pages and components, making information easier to scan and interact with. At the same time, increasing corner radius of the UI makes it feel more approachable, and through a flexible system brings greater consistency to our products, helping related components feel more integrated and visually connected. Together, these refinements simplify the overall interface, reduce unnecessary visual noise, and contribute to a lighter, more modern experience that aligns more closely with the broader Microsoft 365 design ecosystem.
   Increasing roundedness of UI improves the approachability and overall product aesthetics; on the right are examples of how the rounded corners may appear on a page with a group of cards.

What’s Not Changing

While the visual refresh introduces meaningful visual improvements, core SharePoint concepts and workflows remain familiar.

• Your content, structure, and brand are preserved
• Existing site architecture remains unchanged
• Day‑to‑day workflows continue to work unchanged
• There is no impact on existing SPFx extensions or solutions with this change

This update focuses on evolution, not reinvention, so users can benefit from improved clarity and modern visuals while familiar work flows and patterns remain the same. 

Evaluated with Research

Across our research studies, participants consistently favored the Visual Refresh due to the cleaner and more contemporary look and improved labeling and structure.

• A calmer, more modern UI that’s easier to scan. Elevation + neutral theming made the page feel cleaner and it is easier to focus on the content.
• Clearer actions in the command bar. Stronger affordances (like button outlines) make common tasks - edit, undo, save, share - more obvious and easier complete workflows.
• Less guesswork when navigating. Icons and labelling in the app bar reduced friction and participants spent less time hovering and interpreting icons, especially those less familiar with SharePoint.

Overall, the Visual Refresh provides users with a new look they prefer without slowing down their workflow.  

Looking Ahead

The SharePoint Visual Refresh is part of an ongoing journey. We’ll continue refining the experience, learning from customer feedback provided directly in-product, and shipping improvements incrementally, so SharePoint keeps getting better without disrupting how people work.

Familiar workflows will remain in place, now enhanced by improved clarity, consistency, and a more modern feel. We welcome all feedback!

See this post for more information about SharePoint’s exciting next chapter.

Assignments is where the daily work of teaching comes together — creating assignments, collecting student work, giving feedback, and tracking grades. Whether your school uses Microsoft Teams for Education or a learning management system like Canvas or Schoology, Assignments works where you already are. In Teams, it's built into every class team. In your LMS, it's available through the Microsoft 365 LTI® app.

Over the past year, we've added AI tools directly into the assignment workflow. They show up while you're writing instructions, building a rubric, or grading student work. Every AI suggestion is editable, and you decide what stays. AI features in Assignments run within the Microsoft 365 education boundary — student data stays within your tenant and is not used to train AI models.

If you're new to Assignments, this MS Learn module walks you through the basics. It covers the full feature set, including recent AI additions. 

Building an assignment

You start with a title, instructions, and a due date. From there, you shape the assignment to fit what your students need:

• Attach resources — add Word, PowerPoint, PDF, or other files. You control whether students can edit them or just view them. Students can also snap photos of handwritten work and submit them as images.
• Add a quiz — attach a Forms quiz and scores sync back to your gradebook automatically. One of the most-used features in Assignments.
• Add apps — embed EDU apps like MakeCode, Nearpod, or Wakelet directly in the assignment.
• Choose your audience — assign to the full class, a group, or individual students. Group assignments support both group and individual grading.
• Include collaboration tools — add a Whiteboard for brainstorming, group work, or individual submissions.

As you write your instructions, AI suggestions appear below the text box — you can add detail, add hints, add steps, add learning objectives, or align to academic standards (Common Core, NGSS, or your local curriculum framework), each with one click. The suggestions build on what you've already written, so they extend your instructions rather than replacing them. Use one, stack a few, or skip them.

Setting expectations with rubrics and guidelines

Rubrics are one of the most effective ways to show students what good work looks like. You can create a rubric from scratch or generate one from a description and a standard. Enter what the assignment is about, select a standard from the built-in library, and Assignments builds a multi-criteria rubric with proficiency levels. The result is fully editable — rename criteria, adjust descriptions, delete rows, or use the Enhance with AI panel to refine specific sections.

For middle and high school classes where students may use AI tools for their work, you can set Student AI Guidelines directly on the assignment. Four levels — from full AI use to no AI — are visible to students before they start working. You can set a default level and copy your settings across classes with Import Settings.

Practice, review, and check in

Assignments integrates with tools that help students practice, build skills, and reflect — before, during, or after an assignment.

Learning Activities

Learning Activities let you create study materials directly from the documents you've already attached. Click the menu on an attached file and choose Flashcards, Fill in the Blanks, or Matching — the activity is generated from the document's content. Each activity is editable, so you can adjust cards, add hints, or remove items that don't fit.

Learning Accelerators

Learning Accelerators give students personalized practice with real-time feedback:

• Reading Progress — students read aloud, and AI detects fluency patterns like mispronunciations and omissions. You see individual and class-wide progress in Insights.
• Math Progress — AI-generated math problems with auto-grading and progress tracking.
• Search Progress — students practice search and information literacy skills within an assignment, with scaffolds and class-level comparison data.
• Speaker Progress — students practice presenting, and AI coaches them on pace, filler words, and delivery.

You can assign a Reading Progress passage or a Search Progress activity the same way you assign an essay or a project.

Reflect

Reflect brings social-emotional check-ins into the classroom. Students respond to a prompt using emoji-based emotion selectors, and you see class wellbeing trends over time in Insights. You can schedule regular check-ins, run brain breaks, and use the data to spot students who might need extra support — all without leaving Teams.

Grading and giving feedback

When student work comes in, Assignments supports multiple ways to grade and respond:

• Rubric scoring — score each criterion, and students see exactly where they landed across proficiency levels.
• AI-drafted feedback — after you score a rubric, AI drafts a personalized summary in student-friendly language. It highlights what the student did well, where they can improve, and what to focus on next. You review it, edit anything that doesn't sound right, and return it. The draft is a starting point for your own voice.
• Inline markup — highlight, comment, track changes, and use digital ink directly in Word and PowerPoint submissions.
• PDF annotation — mark up PDF submissions.
• Video and audio feedback — record a short video or audio response for a more personal touch. These options give students feedback they can listen to, which is especially helpful for younger learners, students with reading difficulties, or anyone who benefits from hearing tone and emphasis.
• General comments — a text field for overall feedback.

Students can revise and resubmit, and you can return work multiple times. The goal is a draft-feedback-revision loop — students improve their work based on your input, not just a final grade.

You can also configure grading to match your school's policies — weighted categories, letter grades, complete/incomplete, or custom text instead of points.

Built-in accessibility

Immersive Reader is built into Assignments, so students can have any assignment text read aloud, adjust font size and spacing, highlight parts of speech, or translate content into their language — right where they're working. Combined with video and audio feedback, students who struggle with reading or written feedback still get the full experience.

Keeping track across classes

If you teach multiple sections, the cross-class view lets you see all assignments across every class from one place — Upcoming, Ready to grade, Past due, Returned, and Drafts — without switching between teams. You can also view assignments within a single class when you need to focus on one group.

Works in your LMS

If your school runs a learning management system, the Microsoft 365 LTI® app brings Assignments and other Microsoft 365 education tools directly into your LMS through the LTI 1.3 standard. You can create assignments with AI-generated rubrics, collect submissions, and give AI-drafted feedback — all without leaving your LMS.

The Microsoft 365 LTI app works with:

• Canvas by Instructure
• PowerSchool Schoology Learning
• Blackboard by Anthology
• Brightspace by D2L
• Moodle™

Your IT admin handles the initial setup. Once installed, educators and students access Microsoft tools from within the LMS. For setup guidance, see the Microsoft 365 LTI deployment guide.

Getting started

If your school uses Microsoft Teams for Education, Assignments is built in — look for the Assignments tab in any class team. If your school runs an LMS, ask your IT admin about the Microsoft 365 LTI app.

Assignments and its AI features are included with Microsoft 365 Education A1, A3, and A5 licenses at no additional cost.

Helpful Links

Assignments

• Create an assignment in Microsoft Teams
• Organize content, create assignments, and assess learners' understanding in Teams (MS Learn)
• Set Student AI Guidelines on an assignment
• Import Settings in Assignments and Grades

Learning Activities and Learning Accelerators

• Learning Accelerators for Educators (MS Learn) — Reading Progress, Math Progress, Search Progress, Speaker Progress, Reflect, and Insights
• Build social and emotional skills with Reflect (MS Learn)
• Leverage data for action with Insights (MS Learn)

LMS integration

• Microsoft 365 LTI deployment guide

Community

• Join our EDU Insider Program: aka.ms/joinEIP

Have questions or want to share how you're using Assignments in your classroom? Drop them in the comments below — I'd love to hear from you.

Until next time,

Leif Brenne