After years of running AKS in production (and learning the hard way), I wrote a book about the decisions that really matter. 📘 The AKS Book: Every AKS Decision That Will Save You a Cluster Rebuild It’s out now. More info 👇 pixelrobots.co.uk/2026/03/th… #AKS #PlatformEngineering #Kubernetes

We’re building Copilot as a new OS for work that spans every model, every form factor, and every task. Today, we’re announcing our biggest update to Copilot to date, bringing four things together: · Autopilot: proactive and long-running agent built for the enterprise · Code: build apps with Copilot, hosted inside your company’s tenant · Home: Chat + Cowork together · Office: now fully embedded in Copilot (and Copilot embedded in Office, of course!) Plus, you can invoke Copilot in Teams, and we’re introducing Today, a proactive experience that surfaces the most important information from across M365 without needing to ask for it. The way we work is changing and so are our workflows. This update brings AI into that flow, from answering a question, to building an app, to getting work done on your behalf.

KEDA v2.21.0 is out! 🎉 New Azure Cosmos DB Change Feed, ClickHouse & Cloud Spanner scalers; performance improvements; safer service account token auth. Full Changelog: github.com/kedacore/keda/rel… Highlights and upgrade notes: kedify.io/resources/blog/ked… #KEDA #Kubernetes

Your Architecture Diagram is not Your Resilience. A workload can be healthy from an infrastructure perspective and still fail its users, because the dependency that broke was an AI model, not a machine. New blog and interview with Mark Russinovich: msft.it/6011age7U

The deep dive I promised is live: AKS Flex Nodes 🚀 Flex nodes let you attach your own VMs or bare metal to a managed AKS cluster and run them as real nodes, without turning them into a standard node pool. You keep the box, AKS keeps the lifecycle. I built it end to end on a real cluster, so the post covers what actually happens. The reconciliation model, the bring-your-own networking with Unbounded-Net, upgrades that reimage the node not the host, clean removal, and the gotchas that caught me out (including a 409 race on attach and a preview docs gap I filled in). It's preview, so keep it out of production. But if you run at the edge, have bare metal to bring in, or wrestle with data residency, it's well worth a look. 📖 Deep dive: pixelrobots.co.uk/2026/09/ak… #AKS #Azure #Kubernetes #PlatformEngineering

Just put AKS Flex Nodes (preview) through their paces for an upcoming deep-dive blog. 🧪 Spun up a cluster, added a flexpool, scheduled a pod, watched it land. Scale-to-zero + on-demand VM capacity without the usual node pool rigidity. Full write-up soon. #AKS #Azure #Kubernetes

AKS on bare metal now runs on Ubuntu, and this is the bit a lot of people were waiting for 👀 Caught this one through my AKS Docs Tracker. Until now, AKS on bare metal meant validated Azure Local small form factor hardware running Azure Linux. Ubuntu changes the entry point completely. With Ubuntu you bring your own hardware, prep the host, and connect it to Azure Arc. Kubernetes stays AKS-managed, but the host OS lifecycle is yours. Good fit for existing kit rather than buying new validated boxes. Worth knowing: 🔹 Cluster creation is Azure CLI only with Ubuntu, Azure Linux also gets portal, ARM, and Bicep. 🔹 The portal shows read-only status on Ubuntu, so lifecycle work happens through the CLI. 🔹 Both run Kubernetes directly on the host with no hypervisor, and both use Arc. Pick Azure Linux for new validated hardware, pick Ubuntu to reuse your own and manage the OS yourself. 📖 Docs: learn.microsoft.com/en-gb/az… 🔎 Found via my AKS Docs Tracker: pixelrobots.co.uk/aks-docs-t… #AKS #Azure #Kubernetes #PlatformEngineering #BareMetal

Flex Nodes for AKS just landed in preview, and it's a big one 👀 I spotted this through my AKS Docs Tracker before it hit the usual channels, so consider this an early heads up. Flex nodes let you attach your own virtual machines or bare metal hosts to an AKS cluster without moving that compute into a standard node pool. You keep control of the machines, and AKS still handles the node lifecycle through a lightweight open source agent that preps the host, joins it to the cluster, and keeps its config aligned. Why I think this matters: 🔹 Capacity outside your AKS region, handy when data residency rules get in the way. 🔹 Bare metal plus both amd64 and arm64 hosts, so lab and cross-architecture testing gets easier. 🔹 Flexible auth with Azure Arc, VM managed identity, or a service principal. It is not a replacement for standard node pools, it sits alongside them. And it is preview, so I would keep it away from production for now. 📖 Docs: learn.microsoft.com/en-gb/az… 🔎 Found via my AKS Docs Tracker: pixelrobots.co.uk/aks-docs-t… Deep dive blog post coming soon, so keep an eye out. 🚀 #AKS #Azure #Kubernetes #PlatformEngineering #Cloud

From the Blog: GitHub Actions OIDC: Moving to Immutable Subject Claims Link: blog.builtwithcaffeine.cloud… #Community #CloudFamily #BuiltWithCaffeine

I’ve been thinking about what AI agents might mean for Infrastructure as Code. Today, in Azure, we typically describe infrastructure using Bicep, Terraform, ARM, etc. But if AI agents become good enough at translating intent into deterministic infrastructure definitions, do humans still need to author the IaC directly? Could we move towards something like Infrastructure as Markdown (IaM)? Imagine writing: “Deploy this application in UK South. It must be zone resilient, private by default, use managed identity, keep secrets in Key Vault, follow our production security baseline, and stay under £2k/month.” An infrastructure agent could then take that intent, combine it with approved organisational patterns and policies, and generate the Bicep/Terraform. So the flow becomes: Human intent → Markdown → AI agent → IaC → policy/what-if → Azure In that model, IaC doesn’t disappear. It becomes more like an intermediate representation, deterministic, reviewable and auditable, while humans work at a higher level of abstraction. The interesting part for me is whether Markdown could become the actual source of truth. Infrastructure.md with the Bicep or Terraform underneath being generated and maintained by agents. Obviously there are big questions around ambiguity, drift, reproducibility, approvals and how much autonomy an infrastructure agent should ever have. But I’m curious is anyone already experimenting with something like this? Are you using agents to generate IaC from architecture docs, requirements or natural-language infrastructure intent? And could “Infrastructure as Markdown” realistically become a layer above IaC? #Azure #IaC #Bicep #Terraform #AI #AIAgents #DevOps #PlatformEngineering #CloudComputing

How to configure minimum TLS version, elliptic curves and PQC for Istio danielstechblog.io/how-to-co… #Kubernetes #Istio

help us out in CNCF by filling out a new survey on cloud native workload types, we will share the results at #kubecon #cloudnativecon in November! "2026 Cloud Native Workloads Survey" research.net/r/6G5WVBT

When does Kubernetes make sense? It is one of the questions we ask during private Kubernetes courses at LearnKube. My controversial answer is one application.

TIL Unbounded unbounded-cloud.io/

Introducing Agentweaver: an experiment in multi-agent orchestration and runtime enforced workflows on Kubernetes sabbour.me/2026/09/13/introd… #Copilot #Agent #AgenticSDLC

Microsoft Foundry is the best platform for running agents in the enterprise - and one reason is the observability and governance features to keep your agent within the bounds you define 🔐🔎 Long running agents can reason, act, and use tools to help transform so many parts of a business - but doing so responsibly means you need to be able to confidentially answer: What can it access? What can it influence? Where can data go? Can I reconstruct what it's doing? Can I control and budget costs? Check out the demo below to see how Foundry provides a platform to answer all of these questions for you, and gives you a slick agentic developer experience to get it all up and running faster than ever before 🎉🧑‍💻

This article explains what changes when you move from the archived Kubernetes Dashboard to Headlamp, mapping each familiar workflow onto the new UI ➜ ku.bz/v8jGN5MqH

oh I know

Get your .NET MAUI apps ready for Duo today…. Like I did back in 2020 learn.microsoft.com/en-us/do…

KubeBuddy has had a couple of updates recently, and I forgot to post about the last one. The latest release focuses on maintenance and security, keeping the Go CLI and Headlamp plugin dependencies up to date. The bigger update: KubeBuddy now includes 155 total checks, including 146 Kubernetes checks, covering workloads, RBAC, security, networking, storage, events, Prometheus, AKS, and GKE. Recent checks include deprecated Kubernetes APIs, CoreDNS risks, NetworkPolicy gaps, exposed admin interfaces, hostPath risks, ServiceAccount token issues, PDB problems, HPA issues, and StatefulSet validation. Repo: github.com/KubeDeckio/KubeBu… Docs: kubebuddy.kubedeck.io

This case study shows how a team migrated their Spring Boot microservices from Azure App Service to AKS, resulting in a 25% reduction in their infrastructure bill ➜ ku.bz/9kf3cH2K4