By Brian Benz, Paul Yu, Johnson Shi

GPU memory allocation can be tricky to get right. Allocate too little and your AI workload can slow down or fail with an out-of-memory error, often during peak traffic when reliability matters most. Allocate too much and expensive GPU capacity sits unused. Getting an informed view of your workload's actual memory needs from the start helps you avoid both problems.

Kubernetes dashboards can show that GPU memory is under pressure, but they cannot tell you which functions in your AI workload are responsible. Without that detail, teams are left shrinking the workload, moving it to a larger GPU, or restarting it and hoping the problem does not return. That guesswork gets expensive fast.

GPU memory profiling on Azure Kubernetes Service (AKS) provides the evidence to show which functions allocate GPU memory while the workload runs. You can use that visibility to size memory with an informed perspective from the beginning, troubleshoot unexpected pressure, and validate that later adjustments work as intended.

Preview

This preview covers GPU memory allocations, not GPU compute utilization. Your existing GPU metrics tell you that something is going on; the profile gives you specific details on where it is happening.

From guessing to seeing

A GPU memory profile is a record of where allocations occur across your application's function call paths while the workload runs. Instead of showing only how much memory is in use, it connects that memory to the functions that requested it. This gives you a path from the symptom to the code responsible.

In Grafana, the profile is visualized as a flame graph displayed in an icicle-style, top-down layout. Each block represents a function in a call path, and wider bars represent more attributed GPU memory. That helps you identify where to investigate and make targeted changes.

GPU memory profile flame graph showing allocation call paths

You don't add profiling code to your app, and you don't need a sidecar or pod restart to turn it on. Enable profiling on the cluster before deploying your workload, or enable it while an existing workload is running so its CUDA allocation calls can be captured. Profiling won't decide for you that an allocation is wasteful, and it won't tune your workload automatically. It gives you the evidence to decide what your workload needs.

How it works

GPU memory profiling on AKS runs on Inspektor Gadget, an open-source, eBPF-based observability framework for Kubernetes and a CNCF project that Microsoft helps maintain. For GPU profiling, it traces CUDA memory-allocation calls directly from the node, which is why there's no profiling code or sidecar container needed. Enable profiling before deploying the workload, or restart it afterward when you need to capture allocations that occurred during startup.

It's also secure: the Inspektor Gadget project recently completed its first independent security audit; three issues were found and fixed in version 0.51.1.

Results flow into Pyroscope and show up in Grafana as flame graphs rendered in a top-down layout. Pair that with Azure Monitor managed service for Prometheus, and you get memory pressure in one panel and the code path behind it in another.

For more details and to get started, check out the AKS GPU profiling documentation, which covers the prerequisites, extension installation, and wiring up Pyroscope, Grafana, and the dashboard.

A real before and after

A GPU memory flame graph shows which function call paths are responsible for GPU memory allocations. Unlike a distributed trace, it doesn't show operations in chronological order. Instead, it combines many allocation call stacks into a single aggregated view.

The diagram below illustrates the basic idea in the same top-down orientation used by Grafana. Each block represents a function in a call path, stacked blocks show caller-to-callee relationships, and wider blocks represent more attributed GPU memory. The diagram is conceptual, not an actual AKS profile; its colors and example hotspot are illustrative rather than indicators of memory waste.

Conceptual top-down flame graph diagram showing function call paths as stacked bars and bar width as attributed GPU memory

Conceptual flame graph illustration in the same top-down, icicle-style orientation used by Grafana. Read from top to bottom. Wider blocks mean more GPU memory is attributed to that function or call path.

In the Grafana view below, read the graph from top to bottom. Start at the total allocation at the top and follow the function calls downward. Each horizontal bar represents a function. Its width represents the memory attributed to that function and the functions it calls, while horizontal position does not represent time. A tall stack of equally wide bars does not mean each function allocated that memory separately; the same allocation is attributed through the call path.

The AKS documentation includes a vLLM inference example. At the top, the profile shows 65.9 GiB of attributed allocations. One wide branch accounts for 55.1 GiB. Follow that branch downward through the Python and vLLM call stack to find where the memory was allocated.

Expanded Grafana flame graph for a vLLM workload showing GPU memory allocation call stacks

Expanded flame graph from the AKS documentation. Read from top to bottom, following the widest allocation path toward the function with high self memory allocation.

Two terms help you interpret the bars:

• Total: memory attributed to a function, including allocations made by functions it calls.
• Self: memory allocated directly by that function, rather than deeper functions in the recorded stack.

A bar can show 55.1 GiB of total allocation with little or no self allocation because the memory was allocated further down its call path. Keep following the branch downward until you find high self allocation. In this example, that leads to GPUModelRunner._allocate_kv_cache_tensors, which the profile attributes with approximately 55.1 GiB of self allocation. This function creates the KV cache tensors that vLLM uses to reuse previously computed attention keys and values during inference.

The flame graph identifies where memory was allocated, not whether that allocation is unnecessary or the underlying cause of an out-of-memory error. The profile also describes recorded allocations, which are not necessarily the same as the amount of GPU memory simultaneously in use.

A large KV cache isn't automatically a problem; it's doing real work. The real question is whether it's sized for the traffic you actually serve. For a vLLM service, that usually means comparing your configured context length and concurrency against real-world usage, then right-sizing the settings that control them.

Here's a loop to find and fix issues in the flame graph data:

1. See the memory pressure.
2. Find the allocation path in the flame graph.
3. Change one setting, based on what your workload actually needs.
4. Run the same load again, and profile it again.
5. Compare memory, performance, and reliability before you call it done.

What this means for reliability, capacity, and cost

Reliability is the most direct win. When you can trace an out-of-memory failure back to a specific allocation, you fix the actual cause instead of guessing, and you can prove it by profiling again.

Memory efficiency comes next. You might find memory you can safely give back, or confirm the workload genuinely needs what it has. Both answers are useful: one frees up headroom, the other stops you from "optimizing" something that didn't need it.

Capacity and cost are downstream of both. A validated drop in memory use might let you fit more work on the GPUs you already have, or reconsider the GPU size you're requesting; call it responsible resourcing. The profiler isn't a cost calculator and won't resize nodes for you, so you still have to test the change and do the math yourself.

Either way, the decision is now backed by a real allocation path and a repeatable test, not a hunch.

One more example: fine-tuning Stable Diffusion XL on KubeRay

Everything so far has focused on inference, but the same workflow applies to training. The KubeRay GPU profiling demo for AKS with Anyscale spins up an AKS Automatic cluster with GPU Node Autoprovisioning, Inspektor Gadget, Pyroscope, Azure Monitor managed Prometheus, and Azure Managed Grafana.

The demo runs Anyscale's Fine-tuning Stable Diffusion XL with Ray Train template: DreamBooth LoRA fine-tuning across four GPU workers, moving through model loading, training, checkpointing, and image generation.

It's a useful reminder that profiling isn't just for inference. Watch the metrics, find the allocation path, make one change, profile again; the loop works whether you're serving a model or training one.

Try it yourself

Start with the GPU profiling documentation on Microsoft Learn for setup. Once you're running, it's worth exploring the open-source contributions:

• Inspektor Gadget
• Advanced GPU Observability dashboards
• Pyroscope

Preview

GPU memory profiling on AKS is in preview, so there could be breaking changes as the product evolves in response to feedback. Point it at a workload, see what the flame graph can tell you, fix your GPU allocation issues, and share feedback with us on what you find useful and what could be improved.

Ready to put agentic AI to work? Discover how the new GitHub Certified: Agentic AI Business Professional certification can help validate your skills. 

Agentic AI offers business professionals new ways to organize work and automate multistep workflows. With GitHub and GitHub Copilot, business professionals can organize work and build automations while applying human oversight to workflow decisions. 

We're introducing the GitHub Certified: Agentic AI Business Professional certification which validates skills in using GitHub and GitHub Copilot to design agentic workflows and apply human oversight. The certification is earned by passing Exam GH-700: Enhancing Operational Efficiency with Agentic AI, currently available in beta.  

Is this the right certification for you? 

This certification is designed for business professionals, including product managers, program managers, technical project managers, and business analysts, who use GitHub and GitHub Copilot to improve business outcomes. Deep software development or coding experience is not required. 

This certification demonstrates your ability to: 

• Organize and track work using GitHub.
• Produce reliable outputs with GitHub Copilot.
• Build agentic workflows and automations.
• Apply human oversight to review outputs and manage workflow
  decisions. 

You should have experience using generative AI, managing business projects, products, or processes, and producing work that meets business requirements.  

Ready to prove your skills? The first 300 candidates can save 80%

The first 300 people who take Exam GH-700 (beta) on or before November 9 can get 80% off. Enter GH700B8F5 when prompted for payment during registration. This is not a private access code. Seats are first come, first served, and you must take the exam by November 2. The discount is not available in Turkey, Pakistan, India, or China. 

How to prepare 

Get ready for Exam GH-700 (beta): 

• Review the Exam GH-700 (beta) exam page and Exam GH-700 study guide for details and topics covered.
• Connect with Microsoft Training Services Partners in your area for in- person offerings.
• Read Just How Does One Prepare for Beta Exams? for more preparation ideas.

Ready to get started? 

Remember, discounted spots are limited to the first 300 candidates taking Exam GH-700 (beta) on or before November 9. 

Beta exam rescoring begins when the exam goes live, with final results released approximately 10 days later. Read Creating high-quality exams: The path from beta to live for details. 

Stay tuned for general availability in December 2026.   

Additional information 

• Take certification exams from home or the office. Read Online proctored exams: What to expect and how to prepare. 

• Help shape future Microsoft Credentials. Join the Microsoft Worldwide Learning SME Group for Credentials on LinkedIn for beta exam alerts and opportunities to contribute to future learning and assessments. 

 

Keeping up with Windows doesn't mean constantly checking a single website. This guide maps the Windows release information resources available to you in 2026, so you can find the right information at the right time, whether you're checking what's new, investigating an issue, planning an update, or simply trying to stay informed. We'll talk about each resource individually below, but here's a quick decision guide you can reference.

Start with Windows release health

Windows release health is a good place to start when you need authoritative information about the status of a Windows release. Here you can find the rollout status of each release, known and resolved issues, safeguard holds, servicing milestones, and other information that can help you manage your Windows device estate.

Use Windows release health when you need to answer questions such as:

• What are the latest Windows updates and servicing milestones?

• What is the current status of a Windows feature update rollout?

• Where can I find information about a specific version of Windows?

• Is there a known issue affecting this Windows version?

• Has an issue been resolved?

• Is a safeguard hold preventing certain devices from receiving a feature update?

Tip: Bookmark Windows release health so it's easy to find when you need to investigate an issue or check the status of a release.

Understanding each Windows release

Sometimes you don't need to troubleshoot an issue. You simply need to know which Windows releases are currently supported, which build you're running, or when a version is going to reach the end of updates or end of support. Windows release information pages can help you understand the servicing status and update history for supported versions of Windows.

For information about an individual monthly update, turn to Windows release notes. Release notes provide details about a specific update, so you can understand what's included and review information relevant to that release.

Each in-market version of Windows client and Windows Server has its own update history page with a convenient list of all release notes published to date. For example:

If your organization uses hotpatch updates, there are two additional resources you should know about.

• You'll find the Windows 11 hotpatch calendar on the Windows 11 release information page, providing a dedicated place to check the hotpatch release schedule.

• Then you can reference the dedicated hotpatch release notes for information about hotpatch updates.

Together, these resources help you know when hotpatch updates are scheduled and where to find details about the releases themselves.

Both release information pages and release notes help you understand "Which version or build am I looking at?" and "What changed in this particular update?"

Checking for known issues

When something isn't working as expected, Windows release health provides information about known issues that might have a broader impact on devices, functionality, or deployments across supported versions of Windows client and Windows Server.

Known issue information can help you determine whether Microsoft is already investigating an issue, whether a resolution or workaround is available, and whether an issue has been resolved. You can also find information about safeguard holds that might affect the availability of a feature update for some devices.

Consider checking known issues both when you're investigating unexpected behavior and as part of your planning for a Windows feature or monthly update.

Tip: Windows release health is also integrated into the Microsoft 365 admin center, making it easier for IT admins to manage Windows alongside Microsoft 365 products and services. You can proactively view information or sign up for email notifications to have information delivered straight to your inbox.

Don't see Windows release health in the admin center? Check the prerequisites for licensing and role requirements.

Keep up with timely announcements

Not every Windows announcement or advisory warrants a long-form article. That's where the Windows message center comes in.

The public Windows message center provides concise announcements about Windows servicing, lifecycle milestones, known issues, upcoming changes, events, and other information that might require your awareness.

Think of it as a way to quickly answer:

Is there anything happening with Windows that I should know about right now?
For IT admins that have access to the Microsoft 365 admin center, you can find Windows announcements alongside updates on other Microsoft 365 products in the Microsoft 365 message center there. Windows items in the Microsoft 365 message center often contain more technical details on rollout schedules, impact, recommendations, and compliance considerations.

Want to opt in to receive a weekly digest that includes Windows-related updates posted to the Microsoft 365 message center?

1. In the Message center, select Preferences.

2. On the Custom View tab, select the services you want to monitor, including Windows.

3. On the Email tab, select or add the email address that will receive the notifications.

4. Select the box for Send me a weekly digest about services I select and select Windows.

You can also choose to receive same-day email messages for major updates. This is especially useful if you manage Windows across an organization and need to keep track of changes that could affect deployment or support.

Note: You can enter up to two additional email addresses to receive Microsoft 365 message center notifications. Message center weekly digests emails are currently sent in English only.

Follow @WindowsUpdate for updates and practical guidance

If social media is part of how you keep up with operational changes for Windows, follow @WindowsUpdate on X or on Bluesky.

These accounts share timely information about Windows releases, and now, you'll see a broader range of information there, too. @WindowsUpdate is expanding to include practical Windows guidance that can help organizations and other Windows enthusiasts understand important changes and stay informed about updates that might require attention.

That can include:

• Monthly security update availability

• Release notes and related videos

• Release health and known-issue information

• Servicing and lifecycle milestones

• Important changes that might require operational awareness or action

• Guidance and resources for deploying, securing, and managing Windows devices

The goal is simple: make it easier to discover Windows operational updates in a timely manner so you can plan and prepare appropriately.

Tip: @WindowsUpdate isn't just for IT professionals. Developers, power users, and others who want timely information about Windows updates can follow along.

Go deeper with the Windows IT Pro Blog and Microsoft Learn

Announcements tell you what's happening. Sometimes you need to understand why it matters, what to do next, and where to find guidance you can return to over time.

That's where the Windows IT Pro Blog and Windows documentation on Microsoft Learn work together. Look to the Windows IT Pro Blog when you need context around a change, including why it matters, how to plan or prepare, and what your IT team should consider as you manage Windows in your organizations.

For authoritative technical guidance, turn to Microsoft Learn. Technical documentation on Learn reflects the comprehensive product description, prerequisites, and guidance you should use to deploy, configure, manage, and troubleshoot Windows. It's updated as that information evolves. While a blog post captures information and recommendations at the time of publication, the corresponding technical documentation should be your ongoing reference.

Watch and learn

Prefer to see what's changed rather than read about it? Each month, you can watch a short Windows 11 release notes companion video. Get a quick overview of what's in that month's update and a short tip to help you get the most from your Windows experience. For IT admins, Ctrl + Manage offers a closer look at what's changed, why it matters, where it fits, and what you should do next when it comes to the latest and upcoming capabilities in Windows management.

Ask the experts

Sometimes documentation isn't enough because your question depends on your organization's environment, deployment strategy, or management scenario. Community events add an important dimension to your Windows information toolbox: the opportunity to engage directly with the experts.

Community events like Windows Office Hours, Tech Community Live, Technical Takeoff, and Ask Microsoft Anything (AMA) turn Windows communications into a two-way conversation. They give you opportunities not only to learn, but to ask questions, share your scenarios, and get answers. It's your chance to interact with the engineering and product teams building and supporting Windows, Windows 365, Microsoft Intune, and security capabilities.

Sign in to the Tech Community and follow the Windows events page to be notified when new events are announced.

Put your Windows information toolbox to work

With the right resources at your fingertips, you can stay ahead of changes, make informed decisions, and better prepare your organization for what's next. The key is knowing where to look and choosing the resources that best fit how you work.

What would make this toolbox even more useful? Let us know in the comments.

 

---

Continue the conversation. Find best practices. Join the Windows Tech Community.

For updates on new releases, tools, and resources, bookmark the Windows IT Pro Blog and follow @WindowsUpdate on X. For the latest on how Windows is evolving for organizations and the people who support them, follow Windows at Work on LinkedIn.

Files have always been at the heart of how we work. They're where we create, collaborate, and share ideas. And now, with Copilot and agents, they're taking on an even bigger role, providing the knowledge and context that AI needs to understand our work, connect ideas, and help us get things done.

And beyond work, AI is also transforming how we experience our most important memories. Our photos and videos hold the moments that matter most to us. With AI, we're making it easier than ever to find those memories, relive them, and create something new to share with the people we love.

Today at our event, OneDrive + Copilot: Your file system and photo solution for the AI era, we showed how we're bringing that vision to life, with experiences that make your files more useful, your everyday work simpler, and your personal memories easier to enjoy.

We showed Copilot turning files into finished outcomes, a faster and simpler file system across web, desktop, and mobile, easier and more secure sharing and collaboration, and new ways to rediscover, create with, and share your favorite memories.

Let's take a closer look at what's new in OneDrive.

Turning files into outcomes with Copilot in OneDrive

Copilot is embedded in the Microsoft 365 apps where you already work, like Word, Excel, and PowerPoint. Now, with the general availability of Copilot in OneDrive and SharePoint, we’re bringing AI directly to where your files and shared content live. Instead of piecing together information file by file, Copilot can help you understand your content, uncover insights, take action, and create something new.

Here are just a few ways to put Copilot to work:

• NEW - Analyze across files: Select a number of spreadsheets, and ask Copilot to analyze the files, including generating charts and graphs and highlighting trends.
• NEW - Organize your projects: Point Copilot at a crowded folder of reports, research, notes, and plans, and it can sort everything into a structure that makes sense.
• NEW - Create from what you have: Turn current files and past work into something new, like an interactive HTML report, and keep refining it with follow-up prompts.
• NEW - Custom skills in Copilot: Capture a process you repeat as a personal skill, saved as a Markdown file, so Copilot runs it the same way every time.
• NEW - OneDrive Prompt Gallery: Get started with Copilot using ready-to-use prompts that help you explore what’s possible with your files.

And that's just the beginning. Copilot in OneDrive goes well beyond these examples, helping you reason across your content, take action on your files, and get work done simply by describing what you want to accomplish.

Copilot in Action on OneDrive

HTML and Markdown, built for AI-created content

HTML and Markdown are becoming essential formats for Copilot and agents. But creating these files is only the beginning. OneDrive makes it easy to open, edit, and collaborate with them, turning AI-generated content into something you can work with, share, and build on with your team.

• NEW - HTML comes alive in OneDrive: Just like in the Microsoft 365 Copilot experience, you can use Copilot directly in OneDrive to create interactive reports, dashboards, and bring experiences to life. Then edit them directly on the canvas or switch to source/html with a live preview right alongside. Present your HTML output in full screen for an immersive viewing experience. Collaborate with others through anchored comments and live presence, including comments placed directly within slides.
• NEW – Markdown, ready for teamwork: Turn Markdown files into collaborative work. Edit with a familiar rich-text experience while preserving the native .md file underneath. Then refine it together with anchored comments, live presence, real-time coauthoring, and autosave. Collaboration is seamless when you bring people and agents into the same document.

Whether your HTML or Markdown file was created with Copilot or another tool, OneDrive gives you the same rich experience to open, edit, share, and collaborate on it. These files become a natural part of your everyday workflow, right alongside the rest of your content.

Viewing, Editing, Collaborating with HTML and Markdown files in OneDrive

 

Example using Copilot in OneDrive:

So many of us spend hours, sometimes days, pulling together that monthly business report. Gathering spreadsheets, sorting through notes, finding last month's report, and trying to make sense of it all.

Now imagine putting Copilot to work. Ask it to analyze your spreadsheets, uncover trends, and organize the files you need. Then have it bring everything together into a beautiful, interactive HTML report that you can refine and share with your team.

And here's the best part: save the process as a personal skill, and next month's report is just a prompt away.

Actions to try:

• Select several spreadsheets in OneDrive and ask Copilot to compare trends across them.
• Ask Copilot to organize a cluttered project folder.
• Select one or more spreadsheets, your notes stored in a Word document and ask Copilot to create an HTML dashboard/report summarizing all it. Share it and then edit it in real time with your team.
• Ask Copilot to save a workflow as a personal skill
• Watch the full demo deep dives of Copilot in OneDrive and HTML and Markdown in OneDrive.

Stay in flow everywhere: The core file system, reimagined for AI

Work doesn’t sit still. It moves between your personal files, team libraries, desktops, and mobile devices throughout the day. OneDrive helps you stay in the flow by making shared work easier to find, organize, and access wherever you are.

With refreshed library navigation, modern views and filters, improved shortcuts, and expanded sync capabilities, we're making the core file system simpler and more seamless across web, desktop, and mobile.

• NEW - Libraries view: Get to the SharePoint libraries you use most, right from OneDrive. Move between recently visited libraries, Teams, and favorite sites, or open My sites to see every site you're a member or owner of, including ones you haven't visited recently.
• NEW - Search for libraries and lists: Find the right library or list without hunting through sites.
• NEW - Modern views and filters: Filter and navigate the same way everywhere, so finding content feels familiar.
• NEW - Add to OneDrive, with a dedicated folder: Shared folders or libraries that you add now land in a dedicated Shortcuts folder, so they're easy to find.
• NEW - Sync up to one million items on Windows and MacOS: Large libraries sync reliably on Windows and MacOS with as little as 8GB of RAM.
• NEW - Longer file paths: Handle long file paths smarter to reduce sync errors.

Libraries in OneDrive Web

Example using the Libraries view and Add to OneDrive:

You're working across several project sites and a dozen document libraries. In your browser, from OneDrive, you open the Libraries view, access the library you need, filter it the same way you filter your own files, and add the folder you use every day to OneDrive, where it lands in your Shortcuts folder and syncs to your PC or Mac.

Actions to try:

• Open Libraries from the left pane, then browse your recent, Teams, and favorite libraries. Use the filter control to quickly find the exact library you need.
• Add a shared folder or library to OneDrive using the Add shortcut to My files.  Save it to My files or the new Shortcuts folder – your choice!

Built for team flow: Share, collaborate, and stay in control

Sharing and collaboration should be effortless, without compromising security or control. That's why we're making it easier than ever to share files and work together across Microsoft 365, while giving organizations the tools they need to protect and manage their data.

Hero Links are here

Sharing a file has often meant creating multiple links for different people with different permissions. Over time, that can leave you unsure of which link gives what level of access, forcing you to spend extra time ensuring permissions are set up right, and making access harder to manage.

Hero Links, now rolling out, change that, giving you one link per file or folder by default. If you need to change access, you simply update permissions, and the same URL obeys the new permissions. No need for new links! It’s sharing made simple, while keeping you in full control. You can also use the browser link to share the file, with the same permissions applied.

Hero Links in Action

Coming Soon: Expiring hero link access

With hero links rolling out, we know that managing the lifecycle of permissions on your files is also essential to your work.  That’s why we’re developing new, expiring hero link controls.  These controls let you automatically expire your organization’s hero links on a schedule that meets your needs.

For instance, you can configure a policy that requires hero links shared to your entire organization to expire after 90 days. The file’s hero link will work just the way people expect, but when 90 days elapse, the link changes back to the secure default: Only People Added.

Stronger admin controls

As your content grows across OneDrive and SharePoint, admins need simpler ways to secure, govern, archive, and manage it all at scale.  Our latest updates bring you more intelligence and control to content governance, file lifecycle management, and sync administration.

Here’s what’s new:

• SharePoint Admin Agent: Use natural language to assess content across SharePoint and OneDrive, including site cataloging, and take guided action to keep content secure and AI-ready. Learn more in the SharePoint Admin Agent overview.
• File archive in SharePoint document libraries: Archive inactive files with a policy, and use what-if analysis to see the impact before you turn it on.
• NEW - OneDrive Sync management: Exclude specific folders from sync with a folder-level policy, and view accounts, configuration, policy, and status from the command line with the OneDrive Sync Command Line Interface (CLI). Available in private preview. Sign up here

SharePoint Admin Agent

Actions to try:

• Share a file with a Hero Link, then change its permissions without resharing the link.
• Admins: ask the SharePoint Admin Agent, "Which sites are overshared and what are the risk levels for these sites?"
• Watch the full demo deep dive of OneDrive Sync CLI.

More than a place to store photos: An intelligent home for your memories

Our photos and videos capture the people, places, and moments that matter most to us, from life's biggest milestones to the little moments in between. OneDrive has always helped keep those memories safe and accessible, wherever you go.

Now, we're taking the next step. With new experiences across Windows, web, and mobile, OneDrive is becoming a more intelligent and personal home for your photos. From AI-powered Moments that bring forgotten memories back to life, to natural-language search, AI-powered editing, and new ways to create and share, we're helping you do more with the memories you love.

• NEW - Microsoft Photos on Windows: Microsoft’s all-new desktop experience for your photos brings together images and videos from OneDrive, your PC, and iCloud Photos in one unified Gallery. Rediscover memories through Moments, explore with People, and use natural language to find the photos you’re looking for, all through a consistent experience across devices.
• NEW - Copilot in OneDrive photos: Quickly find moments by simply describing them in your own words, then ask Copilot to turn them into something new to enjoy and share. With a prompt like, “See the photos of my kids through the years and turn it into a slideshow”, Copilot will quickly bring years of photos together and create a share-able slideshow from the results.
• NEW - AI-curated and seasonal Moments: Automatically created collections, bringing your best moments back to you.
• NEW - Map view: Browse your photos on the go by their location.
• NEW - Edit with AI: Fix, refine, and reimagine your photos with easy-to-use AI-powered editing tools. From practical corrections like background cleanup to restyling your photos into something new, like crochet or graffiti art, these editing tools help you to easily enhance your photos.
• NEW - Create: Brings a dedicated creation experience to OneDrive, with new ways to use your photos to create personalized, shareable content like collages, holiday cards, and more.

Creating with Copilot in OneDrive Mobile

Example using the Create hub:

Your wedding anniversary is coming up and you want to make something special for your partner. You find some photos you want to include in a card and edit one of them to remove that pesky object in the background. Then you create a collage of photos from your engagement and wedding day and add a “Happy Anniversary” message to it. Spruce it up with some flowers and color and share the image directly with your loved one. Simple tasks that would’ve taken hours of hunting and multiple app switches, all done from one place!

Actions to try:

• Open the Windows Photos app to see your PC, OneDrive, and iCloud photos in one place.
• Search your photos by describing a memory.
• Fix your photos by auto enhancing the colors or removing an object in the background
• Explore your photos on the map in the OneDrive Android app.
• Use Copilot in photos to create a new collage of photos from last year’s holiday season in the OneDrive Android app.
• Create a new social media post using a preset template for Diwali in the OneDrive iOS app.
• Watch the full demo deep dive of Windows Photos app.

The future of OneDrive: Building on what you saw today

We've shared a lot of exciting innovations today, with some available now and others rolling out over the coming months. And we're just getting started.

Looking ahead, we're building on these innovations across Copilot, the core file system, collaboration, and photos and video. Here are a few highlights from the OneDrive roadmap for the coming year.

Copilot: Bringing agents into file collaboration

As Copilot evolves from individual assistance to teamwork with agents, the files we work with remain central to how that collaboration happens.

We're exploring ways to bring agents into familiar file collaboration experiences, making it possible to share a file with an agent, @mention it, and continue working together through comments, just as you would with a colleague.

AutoPilots Working [for You] in OneDrive

Core file system: A home for AI-created content

As more content is created by Copilot and agents, OneDrive will make it easier to find, understand, and manage that work alongside your own. You'll be able to quickly identify files AI helped create, see which agents contributed to them, and understand what changed.

OneDrive, a Natural Home for AI Created Work

File collaboration: Simpler access, stronger security

As people and agents collaborate across more files, managing access becomes increasingly complex. Knowing who can access what, understanding why permissions aren't working, and preventing oversharing are more important than ever.

Future Copilot experiences will make it easier to diagnose and resolve access issues directly from the sharing experience. Copilot will help identify problems such as broken permission inheritance, recommend corrections, and let owners review and apply fixes.

It will also proactively help to uncover oversharing risks, including Anyone links that expose content more broadly than intended.  That way your teams can collaborate with confidence while keeping their information protected.

Copilot Diagnosing and Fixing Sharing Permissions

Photos and video: Easier media workflows for teams

So much of our work relies on photos and video, but finding the right media, keeping it organized, and managing feedback and approvals can be difficult.

We're building new experiences that make it easier for you to find photos and videos by describing their contents, without needing to know filenames or folder locations. Teams will be able to review, comment, and edit together in context, with approvals that stay with the asset, so everyone knows what's ready to use.

Future Support for Commercial Media

Everything we shared today reflects our vision for the next chapter of OneDrive.

The file system is one of computing's most enduring concepts, and in the AI era, it's more important than ever. As people and agents create and collaborate in new ways, they need a durable home for their work and knowledge. That's why we're evolving OneDrive to operate at agent scale, support the formats AI uses naturally, and make the knowledge in your files easier to find, manage, and put to work.

At the same time, we're bringing that same spirit of innovation to your personal photos and videos, helping you rediscover the moments that matter, relive your favorite memories, and create new ways to share them.

That's the future we're building: OneDrive, your file system and photo solution for the AI era.

Resources & next steps

To help you get started and dive deeper:

• Watch the full event, AMA, and demos on YouTube: https://aka.ms/OneDriveEvent/Playlist
• OneDrive adoption resources
• Feature deep dives: OneDrive CLI | Photos App | Copilot in OneDrive | HTML / Markdown
• OneDrive Customer Office Hours
• Sync Up podcast
• Windows Blog
• SharePoint Admin Agent overview

Ready to take action? Bookmark the OneDrive blog at https://aka.ms/OneDriveBlog for the latest tips and feature updates.

There’s been a lot happening with Azure SRE Agent lately, with new guidance, examples, and resources to help teams get more out of the service. Whether you’re thinking through security and architecture, connecting tools like Splunk, or looking for practical ways to improve how your agents work, there’s plenty to dig into.

So, we’ve pulled the latest updates together in one place. Here’s a look at what’s new and what to start.

Understand how Azure SRE Agent handles security and data

Security questions often come first when teams evaluate Azure SRE Agent: What can it access? Where does data go? How are actions controlled? Our updated security overview and data residency guidance answer those questions in one place.

Our latest guidance brings those answers together. It explains how Azure SRE Agent approaches security through execution isolation, secretless credential management, customer separation, encryption, network controls, and logging. It also covers how data such as prompts, responses, conversation history, and resource analysis is stored and processed, including important data residency considerations when choosing a model provider.

Explore the guidance docs: Security overview for Azure SRE Agent, Data residency and privacy in Azure SRE Agent and If you want a secure AI agent, start restricting its environment.

Plan a governed enterprise deployment

Get more from your agent from day one

Creating an agent is the easy part. The real value comes from giving it the right scope, data, and operational context. Our new practical setup guide explains the choices that make the biggest difference. The guidance covers how to start with a focused set of resources, connect code and telemetry, configure incident integrations, and add the runbooks and knowledge your team already relies on. These steps help ground the agent in your environment so it can support real investigations, not just generic questions.

Read the practical guide: Getting the best out of Azure SRE Agent.

Enterprise Adoption Guide

Getting started with Azure SRE Agent is one thing. Figuring out how to expand it across teams and make it part of the way your organization operates is another. In this two-part guide, we share practical lessons for both ends of that journey: how to get your first SRE Agent up and running, and how to build on that foundation as more teams and use cases come on board. The guides walk through resource design, identity, connec?tivity, governance, and integration patterns so platform and architecture teams can build an approach that fits and scale in their environment.

Read the guidance: Part 1: deploying your SRE Agent and Part 2: from one team to enterprise-wide adoption.

Connect operations and development with Azure SRE Agent Canvas in GitHub Copilot

We are also introducing Azure SRE Agent Canvas in GitHub Copilot Apps and Copilot CLI, a new collaborative experience that brings operational findings and context into GitHub Copilot. It gives developers and operators a shared place to review evidence, work through an issue, and decide what to do next.

To help you see how the experience works, we have put together three resources:

• Documentation: Learn how to use Azure SRE Agent Canvas in GitHub Copilot during an investigation. https://aka.ms/sreagent/canvas/doc
• Blog: Read more about how the experience helps operations and development teams work from the same context. https://aka.ms/sreagent/canvas/blog
• Demo: Watch Azure SRE Agent Canvas in GitHub Copilot support an investigation from initial findings through team handoff. https://aka.ms/sreagent/canvas/demo

Azure SRE Agent Canvas in GitHub Copilot brings cloud operations context into a familiar developer workflow, helping teams investigate and resolve issues together without starting over in another tool.

Extend Azure SRE Agent with plugins and response plans

If your Splunk Enterprise environment is private, two new MCP samples can help you connect it to Azure SRE Agent without opening Splunk to inbound internet traffic. Each sample provides a starting point that you can adapt to your own network and production requirements. Both samples could be useful for anyone who uses SRE Agent and Splunk together. Choose the pattern that matches your setup:

• Same-region deployment: Run Azure SRE Agent and the private Splunk workload in the same Azure region and virtual network, with isolated subnets and private MCP traffic. Private Splunk MCP—same region
• Cross-region deployment: Keep Splunk in another Azure region and route private MCP traffic through Global VNet Peering and private DNS. Private Splunk MCP—cross region.

You can now use a private Azure DevOps repo as a plugin marketplace for Azure SRE Agent. This gives your team one place to manage and share plugins and skills built around the tools, processes, and knowledge you already use. The Azure DevOps plugins blog shows how to connect a repo and install those capabilities in Azure SRE Agent.

If your team uses ServiceNow, you can also create more advanced response plans based on the type of incident you are handling. The ServiceNow response plans blog walks through how to set up targeted plans that trigger the right investigation workflow while fitting into the ServiceNow processes your team already has in place.

Where to start

Not sure which resource to open first? Start with the question you are trying to answer:

• Evaluating security or compliance? Start with the security overview, data residency guidance, and Security and Governance white paper.
• Designing for production? Use the Enterprise Architecture Guide to align platform, identity, networking, and governance decisions.
• Configuring a new agent? Follow the practical setup guide to add the context that improves investigation quality.
• Want to connect operational work with developer collaboration? Explore the documentation, blog, and demo for Azure SRE Agent Canvas in GitHub Copilot.
• Connecting with plugins? Choose the right Splunk MCP sample and learn more from the Azure DevOps plugins blogs and ServiceNow responses plan blog.

Additional resources:

• Get started: https://sre.azure.com/
• Documentation: https://learn.microsoft.com/azure/sre-agent/
• Azure SRE Agent on GitHub: https://aka.ms/sreagent/github
• Azure SRE Agent blogs: https://aka.ms/sreagent/blog

Have feedback or a question? Join the discussion in the Azure SRE Agent GitHub repository.

 

 

 

 

 

 

 

As the Internet Archive turns 30, founder Brewster Kahle looks back at the origins of the Archive and considers the future of libraries, digital preservation, and access to knowledge. In this wide-ranging conversation, Kahle explores how libraries can work together to preserve our digital culture, build collections that remain accessible, and make use of new technologies—including AI—without giving up their role as stewards of knowledge. Joining Kahle in conversation is Stephen Wyber of the International Federation of Library Associations and Institutions (IFLA).

This conversation was originally recorded by IFLA as part of its Futures Club series on September 15, 2026. https://www.youtube.com/watch?v=ChQunGkIO9E

Check out all of the Future Knowledge episodes at https://archive.org/details/future-knowledge 

Download audio: https://media.transistor.fm/40820596/9c136bab.mp3