Some of the best quotes and insights I collected from Amsterdam 2026: co-located events (Platform Engineering Days, FluxCon and Agentic Days: MCP + Agents), Kubecon and CloudNativeCon 2026.
The people you are talking to don't speak DORA. Rephrase your metrics to make them understandable to your stakeholders.
A great perspective on the evolution of platform engineering, GitOps+AI done right by having a security-first approach, and great insights into what it means to build platforms for AI in 2026.
Abbey Bangser - From Cloud Native Apps to Cloud Native Platforms
Stefan Prodan - Vibe coding meets GitOps
VIDEO COMING SOON
Max Körbächer - Creating an IDP for AI engineering
Think of developer burnout as a security proxy: the more burned out your devs are, the higher your security risks.
EU Sovereignty - sovereign cloud and what it means for large organizations (from datacenters to chip manufacturing)
Agents in Kubernetes - Open Source as the foundation for AI infrastructure
Platform engineering - It’s all about product thinking. Platforms as marketplaces.
Humans-as-a-platform. Inclusivity is a reliability strategy for human systems. Dev onboarding equals unproductive headcount.
A new way to bridge local developer and target deployment environments. The highlights: a native load balancer in front of your local cluster, smoother cluster suspend/resume, and the possibility to add external nodes (GPUs in the cloud) to your cluster.
A lot of innovation around running AI agents in Kubernetes, agent attestation using SPIFFE/SPIRE, the focus on security, kgateway used as MCP gateway to monitor and control access to MCP servers.
A unified interface for cloud provider resources; nice idea, especially in connection with the EU Cloud Sovereignty Framework and to bridge capabilities of multiple small providers.
AI is crossing team boundaries but can be used as a communication tool. AI agents as a possible Team Topologies extension.
ADLC - AI development life cycle (check out the great talk by Max Körbächer linked above)
AI Factory - the focus is shifting towards building the infrastructure that will support the AI lifecycle
Universal Mesh - A single entry point to all your services across multiple clouds (Istio)
Who is this platform build for? If your platform relies on tribal knowledge to operate, you might want to rethink your model before it fails.
This post talks a lot about platform engineering. Platforms as a marketplace and how to define Day 2 success for Internal Developer Platforms.
When talking about platform engineering, almost everyone highlighted product thinking and product management. We are still defining what kind of marketplaces internal developer platforms are: self-service versus catering versus other types of markets.
One of the most important learning: the Definition of Done is a capability the user can rely on, not just features that are either difficult to access (poor DevEx) or unreliable (flaky platform). More to come on this topic.
If you build in isolation, you end up delivering for a customer that no longer exists.
Waking up every morning at 6:20 to join the Platform café and the Platform engineering community (thanks David Stenglein for organising and sponsoring).
Meeting in a small café with 20/30 people, among speakers and community enthusiasts, to start the day with deep dives into the very questions that are still left unanswered (platform shared responsibility model, day 2 success, digital sovereignty, etc.).
