Top 10 Platform Engineering Takeaways from PlatEngDay & KubeCon London 2025
- Daniel Bryant
- Apr 17
- 5 min read
Updated: Sep 26
Platform Engineering was the hottest topic in the room at KubeCon, even edging out AI from my perspective! Here are my top 10 observations from the event 🎯 :
KubeCon still has a tech- and tool-first mindset 👩💻
There is some "IDP" platform/portal confusion 🤔
People are starting to see portal-first pitfalls 🚧
There are platform-first visibility issues 🏗️
Leadership buy-in challenges are some of the most significant issues 🕴
Practitioner interests included CI/CD for apps, GitOps practices, and fleet management 👨💻
Early interest in running AI/ML workloads on K8s 🤖
Vendor focus was primarily on cost efficiency, resource optimisation, and security 💰
Market uncertainty will impact projects and products 📉
The CNCF community growth is still strong 👥
As usual, I'll offer a massive thank you to all of the organisers, volunteers, attendees, speakers, and sponsors! I had so many amazing chats, and apologies to all the folks I didn't get a chance to connect with (hit me up on email or socials for a Zoom chat!)
1. Tech (and tool)-first mindset 👩💻
The conversations remained deeply technical, with controllers, APIs, pipelines, and provisioning dominating the airwaves. And while that’s a natural reflection of the engineering audience, there’s a clear gap in conversations about platform strategy, organisational alignment, and change management.
If you're building a platform, you need to do more than ship great tech; you also need to map the platform to business goals, bring stakeholders along for the ride, and design with end-users (developers) in mind. Tech solves nothing in isolation.
I talked about this phenomenon and provided some tips to overcome it in my talk, "Platform Engineering for Software Developers and Architects":
And I heard a lot of good things about Camille Fournier's and Ian Nowland's talk (and I gave their recent O'Reilly book, Platform Engineering, a shout-out in my talk!)
2. "IDP" portal and platform confusion 🤔
The acronym “IDP” continues to cause chaos. Are we talking about Internal Developer Platforms or Internal Developer Portals? Spoiler: it depends on who you ask.
Why this matters: Confusion here causes real-world friction, especially when evaluating vendors, defining scope, or deciding whether to build, buy, or blend.
What you can do: Be explicit about your definitions. If you're using both platforms and portals, document how they intersect. And when speaking with vendors or partners, clarify your context early.
3. Portal-first pitfalls 🚧
Teams starting with a portal-first approach are increasingly hitting walls. While portals may look shiny and promise quick wins, they can obscure the complexity of what lies beneath. Templates-as-a-service often break down when real-world use cases and lifecycle operations kick in.
I chatted with multiple teams at the booth who reported "portal fatigue" after launching a glossy interface backed by brittle automation and lacking an easy way for other teams in the organisation to contribute effectively (which we refer to as "platform democracy").
Key lesson: If the underlying platform doesn’t support Day 2 operations—like upgrades, scaling, and security—you’ve built a facade, not a foundation.
Advice: Start with workflows and automation. Build the APIs and interfaces developers need—then surface them via a portal.
Abby and Phill shared the story of how The Access Group focused on platform APIs when kick-starting their latest hackathon series:
4. Platform-first visibility issues 🏗️
On the flip side, teams that lead with infrastructure and API-based platforms are often invisible to leadership and business stakeholders. You’ve built something powerful, but without metrics, dashboards, or developer-facing interfaces, it’s hard to demonstrate value.
Why it’s risky: Platforms need funding, and funding needs visibility. Leadership wants a measurable impact: improved deployment frequency, reduced lead time, and lower support tickets.
What helps: Tie your internal metrics to frameworks like DORA, SPACE, or DevEx. Visualise usage. Show how the platform reduces toil and accelerates delivery.
To learn more, check out this panel, "How Do You Measure Developer Productivity?" with Jennifer Riggins, The New Stack; Cat Morris, Syntasso; Akshaya Aradhya, Oscilar; Laura Tacho, DX; and Helen Greul, Multiverse
5. Leadership buy-in challenges 🕴
This year saw an increase in conversations about how to promote the platform internally. From Total Cost of Ownership (TCO) to Return on Investment (ROI), platform teams are learning to speak the business's language.
What’s changing: Platform teams are being asked to justify budgets and show outcomes. The good news? There are now playbooks and case studies to help.
Pro tip: Don’t just show cost savings—show time-to-market gains, developer satisfaction improvements, and risk reductions.
Emerging pattern: A one-pager with a clear platform vision + usage metrics + anecdotal wins can be a surprisingly effective internal artefact.
6. Practitioner interests 👨💻
Platform engineers and application developers attending KubeCon often live in different realities, but they overlap in interesting ways.
App developers are laser-focused on CI/CD pipelines, build speed, and reliability.
Platform and infra teams are thinking about GitOps, fleet management, and policy enforcement.
GitOps is the bridge that creates shared practices across dev and ops. This reinforces the need for composable platforms where teams can own and iterate on their piece of the puzzle without stepping on each other’s toes.
My colleague Abby and Sebi from Port provided a whistle-stop tour of CRDs for app developers (and others):
And my colleagues, Jake and Cat, explored several related platform-building topics in their talk:
7. AI/ML workloads 🤖
While it's not yet a dominant theme, running AI/ML workloads on Kubernetes is starting to bubble up. Teams are beginning to ask:
How do we package and deploy ML models?
How do we track datasets and inference pipelines?
Where does AI fit into our existing platform workflows?
Platform teams are starting to build reusable scaffolding for AI/ML. Think: model registry integration, GPU scheduling, and secure data access patterns. Expect to see “ML platform” specialisations emerge within larger platform engineering orgs in the next 12–18 months.
Christine Yen, CEO and co-founder of Honeycomb, provided an excellent keynote focusing on observability within this space:
8. Vendor focus 💰
Unsurprisingly, vendor booths were packed with pitches around cost optimisation, resource efficiency, and security. Everyone is trying to reduce spending while staying secure, and platforms are seen as a way to achieve this.
Pro tip for buyers: Be wary of the “silver bullet” slide. Effective platforms require thoughtful design, not just tools. Vet vendors on how well they integrate with your existing workflows and governance.
Bonus: Ask how their product supports platform extensibility and multi-tenancy—critical for scale.
9. Market uncertainty 📉
It’s rare to hear people talking about the stock market at a tech conference, but this year was different. Macroeconomic jitters, layoffs, and shifting priorities were very much part of the vibe.
Longer-term investments, such as platform building, are now under more scrutiny. Teams are being asked to deliver more with less—and prove value faster.
Our booth conversations with attendees beginning their platform building journey focused on keeping the initial platform scope small, showing wins early, and building iteratively. “Thin slice” approaches are more attractive than big-bang platform launches right now.
10. Community Growth 👥
Despite all the uncertainty, the platform engineering community is on fire. The growth of Platform Coffees, two-track Platform Engineering Day events, and a flood of content and meetups shows that momentum is strong.
People are increasingly eager to share hard-won lessons and compare notes. Platform Engineering is no longer a fringe discipline—it’s a career path.
If you’re new to the space, my advice is to join the Slack groups, attend local meetups, or propose a talk to a future KubeCon or Platform Engineering Day. There’s room at the table, and the community is welcoming.
What are your top takeaways? Anything to add or challenge?
Comments