As 2025 draws to a close, platform engineering has reached a remarkable inflection point. What began as conversations about defining the practice has evolved into discussions about excellence, maturity, and the transformative role of AI. This year marked a shift from "what is platform engineering?" to "how do we do it great?" - and the answers are reshaping how organizations build and scale their internal platforms.

TL;DR: Main Insights

• Platform as a product is not optional - it's the fundamental requirement that separates platform engineering from DevOps
• A dedicated platform product manager role has emerged as critical, with 25% of organizations now having this position
• AI and platform engineering have formed a symbiotic relationship, with AI adoption driving platform maturity and vice versa
• 12 core platform capabilities are now considered table stakes for mature teams
• Nearly 90% of enterprises now have internal platforms, surpassing previous predictions by a year

The insights shared here draw from the upcoming State of Platform Engineering report, based on 518 survey responses and hundreds of conversations with practitioners worldwide who are building platforms across regulated and unregulated environments, from startups to large enterprises.

Ricky, Global Practice Area Director for Platform Engineering at Thoughtworks, brings over 15 years of experience designing large-scale, mission-critical systems across government agencies and startups. Sam Barlien organizes the Platform Engineering Community, managing Platform Weekly, co-hosting PlatformCon, and driving community initiatives including the Ambassador program and content channels.

You can watch the full discussion here if you missed it:

The evolution from definition to excellence

Just 52 weeks ago, the platform engineering community was still debating fundamental questions: What is platform engineering? How does it differ from DevOps or SRE? Today, those conversations have matured dramatically. "We are not arguing as much or trying to decide as much about what should be within an engineering platform or developer platform," Ricky explains. "Now we are advancing that and starting to talk about these more mature, more advanced opportunities."

This rapid maturation reflects real-world adoption. Organizations have moved beyond theoretical discussions to practical implementation, identifying specific capabilities and patterns that separate successful platform teams from those still struggling to gain traction.

The 12 core platform capabilities

Through extensive work with customers globally, Thoughtworks has identified 12 platform engineering capabilities that represent table stakes for mature teams. Two capabilities are binary requirements - without them, you're not doing platform engineering at all:

• Platform product thinking: Treating your platform like a product, complete with user research, roadmaps, and continuous improvement
• Enterprise architecture alignment: Creating golden paths and paved roads that connect platform capabilities to broader organizational goals

The remaining capabilities form the technical and operational foundation:

• Infrastructure automation
• CI/CD pipeline management
• Testing
• Security and compliance
• Observability
• Containerization and cloud-native practices
• API management
• Cost management
• Governance and controls
• Developer experience

"We think that those things are kind of like the core tenants or core capabilities of any of the good platforms that are out there," Ricky notes. "The more successful teams are providing those capabilities to their engineering teams and to the business to actually achieve business outcomes."

Platform as a product: The non-negotiable foundation

Perhaps the most critical insight from 2025 is the absolute necessity of platform as a product thinking. This isn't a differentiator between good and great platform engineering - it's the dividing line between platform engineering and everything else.

"If you are not doing platform as a product, you're not really yet doing platform engineering," Sam emphasizes. This concept traces back to Evan Bottcher's seminal 2018 paper on platform engineering, where product thinking was already identified as fundamental.

But what does treating a platform like a product actually mean? Ricky offers a practical framework: "Think about your platform like it's a SaaS product." This means:

• Having a dedicated product manager
• Creating comprehensive documentation
• Providing customer success support
• Running marketing campaigns for new features
• Conducting user research and surveys
• Building communities around the platform
• Publishing release notes
• Thinking in terms of continuous improvement rather than one-and-done projects

"If I was trying to get my platform consumed and used by a thousand engineers at a very large or midsized enterprise company, what would it take to do all of those things?" Ricky asks. "I have to do surveys. I have to do interactions with those devs. I have to build communities."

The rise of the platform product manager

One of the most significant developments in 2025 has been the emergence of the platform product manager as a distinct, critical role. Survey data shows that 25% of organizations now have a dedicated platform product manager - and the most successful platform teams universally include this position.

"I think you need a platform product manager to navigate the different touch points and to collect this information to actually drive the team," Ricky explains. This person focuses full-time on understanding developer needs, prioritizing features, avoiding backlog coupling, and determining the right interfaces for platform consumption.

However, a critical anti-pattern has emerged: converting project managers into platform product managers without proper enablement and training. "Time and time again I see some of these teams struggling to get from that level three of maturity to a four or five," Ricky notes. "It's generally that you've just taken a project manager and you've converted them to a product manager on the platform side but you've not given them the skills and enablement to actually make sure that they're able to do that job."

For leaders considering this role, the recommendation is clear: dedicate your best product management talent to internal platforms. While this may seem counterintuitive - pulling talent from external-facing products - the ROI is substantial. A skilled platform product manager can quantify platform value, measure impact, and drive adoption in ways that directly benefit all product teams.

AI and platform engineering: A symbiotic relationship

The intersection of AI and platform engineering defined much of 2025's evolution. What started as tentative exploration has become a fundamental driver of platform adoption and maturity.

The relationship works in both directions. First, AI capabilities are accelerating existing engineering practices. "We are seeing more and more clients embed AI agents and agentic workflows around maturing and improving software engineering practices," Ricky explains. Organizations are using AI to improve testing coverage, enhance security practices, and accelerate various parts of the software delivery lifecycle.

Second, platform teams are extending infrastructure to support new AI and ML workloads. This includes providing the right compute resources - GPUs, TPUs, specialized clusters - along with appropriate guardrails, observability, and governance.

The DORA report revealed that nearly 90% of enterprises now have internal platforms, surpassing Gartner's 2026 prediction of 80% a full year early. This acceleration is directly tied to AI adoption. "Initially there was this idea that I could just give my developers a bunch of coding tools and then those coding tools would just improve outcomes," Ricky recalls. "We quickly hit this inflection point of, oh, either they don't want to use it or they're not getting the best outcomes."

Organizations realized that simply providing AI coding assistants wasn't enough. The rest of the software delivery lifecycle - CI/CD, testing, deployment, observability - needed to mature to handle the increased code output. This drove investment in platform engineering capabilities.

Conversely, the desire to use AI tools to solve developers' hardest problems required platform product thinking to identify what those problems actually were. "Even in the Dora report, surprisingly, they had an entire section on value stream mapping," Ricky notes. "That's not surprising to us because when we go in and do a lot of these assessments, we do an end-to-end value stream mapping to identify where the pain points are so that we can actually solve those problems."

The expanding scope of platform engineering

Another defining trend of 2025 has been platform engineering's expansion into adjacent domains. Observability, security, and data engineering have increasingly become part of the platform conversation.

"We have seen in the community an absolutely massive increase in the number of observability-focused titles, data-focused titles, security, FinOps, consuming the content, writing articles, becoming ambassadors," Sam observes. These domains, while always part of the equation, are now explicitly recognized as core platform capabilities rather than separate concerns.

This expansion reflects a maturing understanding of what platforms must provide. Security isn't bolted on - it's designed into golden paths. Observability isn't an afterthought - it's a fundamental capability that platforms expose to developers. Cost management isn't just a finance concern - it's a platform feature that helps teams make informed decisions.

Looking ahead: Agentic platforms and beyond

As 2025 closes, attention is turning to what 2026 might bring. The consensus points toward "agentic developer platforms" - platforms that leverage AI agents throughout the software delivery lifecycle.

"I think there's going to be this convergence of the infrastructure that developers, both AI and ML and all of these personas, need to deliver new AI capabilities," Ricky predicts. This includes not just large language models but the full spectrum of AI techniques: machine learning, predictive analytics, prescriptive analytics, and more.

The key questions for 2026 revolve around interfaces and integration. Should developers consume AI agents through AWS Q for Dev? Code Gemini? CLI tools? IDE plugins? An agentic-powered developer portal? "I think we're still trying to figure out those things and the infrastructure that's necessary to support those things," Ricky notes.

What remains constant, however, are the fundamentals. "Your agentic platform still needs a product mindset, still needs a platform product manager, still needs those fundamentals," Sam emphasizes. "Don't think AI is going to save the day. You still need to get all those fundamentals down first."

If you enjoyed this, find here more great insights and events from our Platform Engineering Community.

If you want to dive deeper, explore our instructor-led Platform Engineering Certified Professional course and connect with peers from large-scale enterprises who are driving platform engineering initiatives.

Key takeaways

• Platform as a product is mandatory, not optional: If you're not treating your platform like a SaaS product - with user research, documentation, customer success, and continuous improvement - you're not doing platform engineering yet. This is the fundamental dividing line.
• Invest in a dedicated platform product manager: The most successful platform teams have someone focused full-time on understanding developer needs, prioritizing features, and driving adoption. Don't just convert a project manager without proper training and enablement.
• AI and platforms need each other: AI adoption is driving platform maturity as organizations realize they need robust platforms to handle AI-generated code and workloads. Simultaneously, platforms are extending to provide AI infrastructure and capabilities. This symbiotic relationship will only deepen.‍
• Master the core capabilities before chasing trends: Whether you're building for AI, observability, or any other emerging domain, the 12 core platform capabilities remain table stakes. Focus on infrastructure automation, CI/CD, security, observability, and developer experience before adding advanced features.