Kubernetes v1.37 brings 67 enhancements: the kubelet can finally run rootless, gang scheduling supports AI/ML workloads, and pod certificates become a native cluster feature. An overview of the key updates for operators.
Software supply chain attacks cost $80.6 billion annually. How SBOMs, Sigstore keyless signing, SLSA build integrity, and Kubernetes admission controllers build multi-layered protection – and why the EU Cyber Resilience Act 2026 becomes the driver.
localreview is a Claude Code plugin for local AI code reviews. Unlike cloud solutions, your code never leaves your computer – the review runs via oMLX on the MacBook, with LM Studio support planned for the future.
From passive pipeline to active agent: How Self-RAG, CRAG, and Adaptive RAG are revolutionizing retrieval-augmented generation in 2026—with real costs, latencies, and production stacks.
Andrej Karpathy's LLM-Wiki approach from April 2026 challenges the way AI manages knowledge: instead of fragments from RAG systems, it relies on a persistent, LLM-maintained wiki in Markdown. An introduction to architecture, operations, and potential.
Local PII filters remove sensitive data beforehand, the cloud LLM processes anonymized content – and locally, results are enriched with original data again. A hybrid architecture for GDPR-compliant AI.
Three open-source projects compete in 2026 for efficient GPU distribution in Kubernetes clusters. A comparison of the approaches of Kueue, Volcano, and KAI Scheduler – including a decision matrix for practical use.
In 2026, Backstage has evolved from a developer portal to a context hub for AI agents. With v1.51.0, AiResource and MCP Server were introduced as new entity types—the software catalog becomes a structured knowledge base for AI workflows.
AI agents need knowledge sources. Obsidian vaults can provide these via the Model Context Protocol (MCP). An overview of the most important MCP servers, setup options, and practical workflows in 2026.
ArgoCD or Flux? Both GitOps tools are CNCF-graduated, but their architectures could not be more different. A comparison for platform teams.
The Model Context Protocol (MCP) has evolved from an Anthropic initiative into a widely supported industry standard. The 2026-07-28 revision finally makes it enterprise-ready – with a stateless core, MRTR, and a formal extensions framework.
A comparison of LangGraph, CrewAI, and AutoGen: which multi-agent framework is best suited for which task, and what has changed in 2026.
How ArgoCD masters the complexity of Kubernetes deployments in a multi-cluster environment – with ApplicationSets, hub-and-spoke architecture, and a focus on security.
Since the license fork in 2023, OpenTofu and Terraform have evolved into independent platforms. A comparison of features, governance models, and decision criteria for 2026.
Platform Engineering hits the 80% adoption mark in 2026. What’s behind the hype, which roles are emerging, and why business metrics are replacing DORA benchmarks.
Phi-4, Gemma 4, Qwen 3.5: In 2026, small language models achieve quality parity with large models—at a fraction of the cost and fully on-premises. An overview of the key SLM families and their enterprise benefits.
Kubernetes 1.37 was released on August 26, 2026. Three breaking changes – cgroup-v1 hard stop, IPVS deprecation, and static pod secret prohibition – require concrete actions before upgrading. All background information and the upgrade checklist.
GraphRAG combines knowledge graphs with retrieval-augmented generation and delivers measurably better results for complex enterprise queries. LazyGraphRAG drastically reduces indexing costs – an overview of the state of the art in 2026.
In 2026, AI agents will handle the initial analysis of incidents in Kubernetes environments. An overview of the tools, architectures, and pitfalls of the new generation of AIOps.
Backstage dominates the internal developer portal market with an 89 percent share — yet only ten percent of developers actually use it. An analysis of the paradox and its causes.
Most RAG systems fail not because of the LLM, but due to invisible errors in the retrieval pipeline. An overview of the key metrics, observability levels, and strategies for reliable production operations.
ListenerSet, TLSRoute, and four other features are stable in the standard channel – Gateway API v1.5 is the project's largest release yet and makes multi-tenant networking in Kubernetes production-ready.
ArgoCD 3.4 is here—with pause reconciliation at the cluster level, Helm wildcard globs, a new Teams integration, and improved UI filtering. An overview of the key features for platform engineers.
eBPF and Cilium have established themselves as the backbone of modern Kubernetes networking in 2026. This article explains why the kernel-native stack makes sidecars obsolete and how Cilium combines CNI, Service Mesh, and Observability.
Deployment frequency, lead time, and change failure rate – the DORA metrics are the gold standard for DevOps teams. But by 2026, it's clear: AI-generated code massively skews these measurements. Why elite teams now need to rethink their approach.
In 2026, Claude Code, Cursor, and Codex CLI are vying for dominance in AI-assisted software development. A comparison of the tools, their strengths, and costs.
Andrej Karpathy's LLM Wiki is not a product, but a paradigm shift: Instead of just searching documents, LLMs permanently compile knowledge into a structured, linked Markdown wiki. An overview of architecture, functionality, and current community implementations.
Obsidian doesn't have its own AI feature in 2026 – but the plugin ecosystem is booming. Copilot V4 brings AI agents directly into the vault, Smart Connections enables semantic search, and with Ollama everything runs locally. An overview of the current state of affairs.
How companies can build privacy-compliant AI pipelines using Ollama, llama.cpp, and open-source RAG frameworks – without cloud APIs and without sacrificing data sovereignty.
With MCP integration, Actions Registry, and AI-powered plugins, Backstage 2026 positions itself as a central interface between developers and AI agents. An overview of the innovations and the controversies.
Dynamic Resource Allocation (DRA) has been stable since Kubernetes 1.34/1.35 and is revolutionizing GPU allocation for AI workloads. Instead of simple device counting, DRA enables attribute-based allocation, device sharing, and time slicing — an overview of the new API and its production readiness.
AI applications are often black boxes—until they make mistakes. In 2026, OpenTelemetry, with its new GenAI semantic conventions, brings transparency to LLM calls, agents, and AI pipelines, establishing observability as the standard for AI workloads.
The Model Context Protocol has evolved from an Anthropic experiment to an industry standard in 18 months. An overview of architecture, enterprise adoption, and the new 2026-07-28 specification.
LangGraph, CrewAI, AutoGen/AG2, and Temporal have reached production readiness in 2026. An overview of the multi-agent orchestration frameworks that bring enterprise AI projects from prototype to production.
Presidio is an open-source framework for detecting and anonymizing personally identifiable information (PII) in text and images. The article shows how it can be placed before LLM pipelines to minimize GDPR risks.
The official Obsidian Web Clipper turns web pages into structured Markdown notes and, combined with AI workflows, forms the foundation of a personal knowledge base. An overview of the tool, its AI integration, and the modernized ecosystem of plugins in 2026.
Argo CD 3.5 is here: internal mTLS encryption, Git commit signature verification, a native ApplicationSet UI, and Helm 4 support — an overview of the key new features of the leading GitOps tool for Kubernetes.
On August 26, 2026, Kubernetes 1.37 will be released – with long-awaited stabilizations, a milestone for node security, and several farewells to legacy systems. An overview of the most important changes.
By 2026, platform engineering has become the defining evolution of DevOps: Internal Developer Platforms, Backstage, Crossplane, advanced GitOps, and AI-powered operations are transforming how organizations deliver software.
How GraphRAG with knowledge graphs overcomes the limits of classic vector-based RAG systems—and which open-source tools will drive the breakthrough in 2026.
ArgoCD Image Updater and Argo Rollouts make ArgoCD 2026 a fully automated GitOps platform – from the image registry to metric-based canary rollouts.
Backstage dominates the IDP market with around 89 percent. The CNCF documentation, the new backend system, Golden Paths, and AI integration mark the current state of the leading open-source developer portal.
In 2026, Karpenter has evolved from an AWS experiment to a serious candidate for node provisioning. The article compares Karpenter with the Cluster Autoscaler, highlights advantages in speed, bin-packing, and spot usage, and helps you decide on the right autoscaler.
Istio Ambient Mode, Linkerd’s simplicity, and Cilium’s eBPF – the service mesh landscape in 2026 is being reshaped. Why Ambient Mode cuts resource consumption by 90% and how platform teams benefit.
Ingress-NGINX was retired in March 2026 – leaving a critical component of many clusters without security updates. An overview of the acute situation, the Gateway API as successor, and five security findings that truly matter in 2026.
Pure vector search hits its limits with error codes, product SKUs, and API names. Hybrid Search with BM25 + Dense Embeddings + Reranking is the 2026 production standard. This article shows architecture, fusion methods, and reranking.
57% are using AI agents in production, but quality and safety remain the biggest hurdles. An overview of evaluation frameworks, benchmarks, and governance strategies for operating autonomous agents.
SLMs are no longer a compromise solution in 2026. An overview of the strongest models, hardware requirements, and use cases—from Phi-4 to Qwen3 to Gemma 3.
Andrej Karpathy's LLM Wiki Pattern promises a paradigm shift: Instead of extracting knowledge from raw documents anew with each query, an AI agent compiles it once into a persistent, linked Markdown wiki. An overview of architecture, tools, and limitations.
ArgoCD and Flux are the two dominant GitOps platforms in 2026. This comparison shows which approach suits which team—from architecture and multi-cluster management to progressive delivery.
The Backstage Scaffolder is the most important tool for platform teams looking to implement developer self-service. An overview of the v1beta3 format, Golden Paths, and custom actions.
In 2026, local RAG systems for Obsidian have made a leap forward: knowledge graphs, MCP integration, and hybrid search are replacing pure vector search. An overview of the current stack and the most promising approaches.
Kyverno achieved CNCF graduated status in March 2026. A comparison of the two leading policy engines for Kubernetes — Kyverno vs. OPA Gatekeeper — and why policy-as-code becomes an operational necessity in 2026.
OpenTelemetry (OTel) has achieved the highest CNCF maturity level. The graduation in May 2026 marks the definitive breakthrough as a vendor-neutral standard for metrics, logs, and traces – and comes at exactly the right time for AI observability.
By 2026, the focus has shifted from individual AI assistants to coordinated multi-agent systems. An overview of architecture patterns, the consolidated framework landscape, and the new communication protocols A2A and MCP.
The market for AI-powered development tools has consolidated in 2026. Three tools dominate—Cursor, Claude Code, and GitHub Copilot—each with different philosophies and price points. An overview of the current landscape.
AI workloads are establishing Kubernetes as the production standard. DRA, Kueue, KubeRay, and vLLM solve old GPU scheduling problems and make container orchestration the top choice for training and inference.
With version 2026-07-28, MCP is fundamentally revamped: stateless, scalable, and enterprise-ready. An overview of the most important innovations in the protocol that connects AI agents with external tools.
In 2026, open AI models have caught up with their closed competitors in many areas. GLM-5.2, Kimi K3, DeepSeek V4, and Gemma 4 show that open-source AI is production-ready.
DevOps is evolving rapidly in 2026: Platform Engineering relieves developers, AIOps automates operations, and DevSecOps integrates security directly into the pipeline. An overview of the three key trends.
Andrej Karpathy's LLM-Wiki pattern adds a new dimension to AI-powered knowledge management: Instead of searching anew for every query, an LLM builds a persistent, linked Markdown knowledge base – and keeps it up to date autonomously.
Backstage v1.53.0, Spotify Portal GA, Red Hat Developer Hub with dynamic plugins, and the New Frontend System — the open-source framework for internal developer platforms is evolving rapidly. An overview of the key new features in 2025/2026.
AppGrid is a grid-based application launcher for KDE Plasma 6, offering a modern, searchable alternative to Kickoff and Kicker. Two variants, KRunner integration, and extensive configuration in an open-source Plasmoid.
Argo CD has established itself as the de facto standard for GitOps on Kubernetes. This article highlights the key innovations: Argo CD 3.0, ApplicationSets, Argo Rollouts, CNCF graduation, and security improvements.
With 70 enhancements, Kubernetes v1.36 sets new standards: User Namespaces reach general availability, AI workloads gain improved GPU scheduling capabilities, and API scalability is boosted via Sharded Watch Streams.
Retrieval-Augmented Generation has fundamentally changed over the past two years: Agentic RAG, Microsoft's GraphRAG, and intelligent guardrails transform the once-simple chunk-and-search pipeline into a flexible, agentic architecture. An overview of the most important developments.
AI agents can act independently, use tools, and pursue complex goals. An overview of LangGraph, CrewAI, AutoGen and Co. – and why A2A is the game changer.
Obsidian plugins enable the integration of AI into note-taking systems. Semantic search, local LLMs, and automated knowledge generation transform the way we organize information.