Top Kubernetes Tools from KubeCon EU 2025: Tech Stack Audit & Emerging Projects

This blog is part of our KubeCon EU 2025 series, we recommend reading the rest of the posts in the series:

• The LiveWyer Guide to London for KubeCon attendees
• Top Kubernetes Tools from KubeCon EU 2025: Tech Stack Audit & Emerging Projects
• KubeCon + CloudNativeCon Europe 2025: 5 Key Takeaways

Kubecon EU 2025 is a wrap!

KubeCon EU 2025 has come and gone, but its impact will be felt throughout the cloud native community for months to come. This year’s event made one thing abundantly clear: the Kubernetes ecosystem is evolving rapidly, and smart tools are emerging to help teams navigate increasing complexity with greater confidence.

We’ve pulled together a short list of some of the standout tools generating buzz at the conference. Some are CNCF projects climbing the graduation ladder; others are newer entrants making waves with clever solutions to old problems, but all were featured in at least one live demo, talk, or lightning talk. Here’s your post-KubeCon audit of what’s worth exploring.

OpenFeature: Standardised, Pluggable Feature Flags

Feature flags are central to modern application delivery, but implementation is often inconsistent across teams and providers. OpenFeature, now an Incubating project under the CNCF, addresses this by offering a vendor-neutral API and SDKs for consistent feature flag usage across environments.

OpenFeature separates flag logic from your codebase and allows seamless switching between backends such as LaunchDarkly, Flagsmith, or open source options. It supports context-aware evaluation, dynamic updates, and integrations with observability platforms. If you’re building internal platforms or working in a multi-provider environment, this gives you the flexibility you didn’t know you needed.

Kubefleet: Multi-Tenant Kubernetes, Simplified

KubeFleet may not yet be a CNCF project, but it drew strong interest at KubeCon for its elegant approach to managing multi-tenant Kubernetes environments. Instead of relying solely on namespaces and manual policy enforcement, KubeFleet introduces a controller-driven architecture that automates tenant provisioning, isolation, and lifecycle management.

It uses custom resource definitions (CRDs) to define tenant blueprints, integrates with policy engines like Kyverno or OPA, and supports dynamic resource scaling. For organisations running SaaS platforms or internal developer portals, KubeFleet makes it feasible to manage multiple tenants in a shared cluster, securely and efficiently.

K8sGPT: AI-Powered Cluster Troubleshooting

Troubleshooting Kubernetes issues can often feel like deciphering an ancient script. K8sGPT, while not (yet) a CNCF project, applies large language models (LLMs) to bring clarity and speed to incident response. It analyses cluster resources, including events, pod logs, and configurations, and returns human-readable explanations, often paired with actionable remediation advice.

Under the bonnet, it combines built-in analysers with GPT-powered summarisation. You can run it via CLI (k8sgpt analyze) or integrate it with your observability tools. For platform and SRE teams, it’s proving a useful addition to the debugging toolkit, especially when triaging unfamiliar issues in complex environments.

KubeStellar: Distributed Cluster Coordination for the Edge

Edge and hybrid environments come with unique challenges, from unreliable connectivity to data sovereignty. KubeStellar, currently in the CNCF Sandbox, provides a way to manage fleets of Kubernetes clusters from a central control plane, even when those clusters are intermittently connected.

It builds on the Kubernetes Control Plane (KCP) project and introduces concepts like sync targets and resource exports to allow workload placement and policy propagation across disconnected clusters. With declarative sync, edge clusters remain autonomous yet manageable. It’s particularly suited for use cases in retail, manufacturing, or logistics, where infrastructure is distributed and latency matters.

Kueue: Queueing and Scheduling for Batch and AI Workloads

Traditional Kubernetes scheduling isn’t well-suited for batch workloads, machine learning pipelines, or compute-heavy jobs. Kueue, now an Incubating CNCF project, enhances the default scheduler by introducing a queueing system that respects resource availability, job priorities, and quota constraints.

It wraps Kubernetes-native Job and Pod resources into Workloads that can be queued until resources are ready. Features like cohort scheduling and quota borrowing allow for higher cluster utilisation and more predictable job completion. For teams working with GPU-bound tasks or bursty workloads, Kueue brings control and efficiency where standard scheduling falls short.

Final Thoughts: What’s Worth Exploring?

Kubernetes itself may be stabilising, but the ecosystem around it continues to innovate at speed. Tools like OpenFeature, Kueue, and KubeStellar aren’t just nice-to-haves, they represent the next step in making Kubernetes platforms smarter, more flexible, and production-ready at scale.

Curious to explore further? Most of these projects are open source and actively welcoming contributors. Whether you want to trial one in a dev environment, ask questions in their community channels, or get involved in shaping the roadmap, now’s an excellent time to dive in.

This blog is part of our KubeCon EU 2025 series, we recommend reading the rest of the posts in the series:

• The LiveWyer Guide to London for KubeCon attendees
• Top Kubernetes Tools from KubeCon EU 2025: Tech Stack Audit & Emerging Projects
• KubeCon + CloudNativeCon Europe 2025: 5 Key Takeaways