The five decisions that matter

  • Do you need a platform? Usually only when multiple teams are working to solve the same problems. → Section 3

  • What do you deliver first? The golden path for the most common task. → Sections 5 and 9

  • Build, assemble, or purchase? → Section 7

  • Who builds and manages it? A platform team with a product owner. → Section 8

  • How do you know if it's working? Measure adoption, turnaround time, and satisfaction. → Section 10

1. What is platform engineering?

Platform engineering involves designing, building, and managing an Internal Developer Platform (IDP): a set of tools, services, and conventions that development teams use on a self-service basis. A team that wants a new service selects a template and is provided with a repository, pipeline, environment, monitoring, and security settings. The team deploys the service without having to submit a ticket to another team.

Gartner predicted that by 2026, 80% of large software engineering organizations would have a platform team, up from 45% in 2022.

Key concepts

  • Internal Developer Platform (IDP): everything the platform team provides to development teams, from infrastructure to pipelines.

  • Developer portal: the platform's front door, featuring a service catalog, documentation, and self-service options. Backstage is the best-known example.

  • Golden path: the recommended, automated route for a common task, such as starting a new service.

  • Guardrails: rules that the platform automatically enforces, such as security policies and cost limits.

  • Platform as a product: The platform has users (developers), a roadmap, and a product owner, just like software for customers.

  • Cognitive load: everything a developer needs to know beyond the application itself. A platform reduces that load.

2. Platform Engineering, DevOps, and SRE

 

Concept What it is Relationship to Platform Engineering
DevOps Culture and way of working: teams build and run their own software Platform engineering makes DevOps scalable, so that not every team needs to have all the infrastructure knowledge itself
SRE Production reliability, with service level objectives and error budgets SRE practices are often built into the platform: monitoring, alerting, and incident response processes
Platform Engineering An in-house product that makes the work of development teams easier and safer Builds on DevOps and SRE

You can also read about the relationship between DevOps and platform engineering in the blog post “Platform Engineering: The Future of DevOps?”

3. When is a platform worth it, and when isn't it?

It pays off when:

  • multiple teams are repeatedly solving the same infrastructure tasks;
  • developers spend a lot of time on pipelines, environments, and tickets;
  • pipelines, security settings, and tools vary by team;
  • it takes a long time for a new developer to get their first change into production;
  • during audits, you must demonstrate that every release meets the same requirements.

It usually isn't worth it if:

  • you have one or two teams with a simple stack;
  • you do not have the capacity to maintain the platform as a product, either on your own or with a partner;
  • a managed PaaS or container service from your cloud provider already does everything your teams need.

You can find out how to make the business case work in *Invent the Feature, Not the Platform*.

4. What does an Internal Developer Platform include?

The portal is the front door; the platform is everything behind it. A developer sees the portal, but benefits most from the automation underneath it.

↑ What the developer sees Examples
Developer Portal and Service Catalog Backstage, Port, Cortex
Templates and golden paths Scaffolding for services, pipelines, and infrastructure
CI/CD and GitOps GitLab CI, GitHub Actions, Argo CD, Flux
Infrastructure and container orchestration Kubernetes, Terraform or OpenTofu, Crossplane
Security and Policies Single sign-on, secrets management, image scanning, Kyverno or OPA Gatekeeper
Observability Prometheus, Grafana, OpenTelemetry, Elastic, Dynatrace
Costs Tags, cost allocation by team, OpenCost

↓ The automation below

5. Golden paths and guardrails

A "golden path" is the route recommended and fully automated by the platform team. For example, a golden path for a new web service provides:

  • a repository with a standard structure;
  • a pipeline that includes tests, security scans, and a container image;
  • a deployment to the test environment via GitOps;
  • a dashboard and alerts;
  • tags for cost allocation.

Guardrails enforce what should always be the rule: no containers running as root, mandatory resource limits, images only from trusted registries, and encrypted storage. A "golden path" is a recommendation: teams may deviate from it, but they are then responsible for taking on the additional burden themselves. The more appealing the path, the less likely teams are to deviate from it.

For a microservices architecture, the blog post "Platform Engineering for Microservices" explores this topic in more detail.

6. Portal or platform?

Many organizations start with a developer portal like Backstage. A portal without effective automation behind it becomes nothing more than a catalog of links and tickets. Therefore, start with the golden path and add the portal once there’s enough to show. Backstage is open source and flexible, but requires in-house development resources to maintain. Commercial portals like Port and Cortex can be set up more quickly, though they offer less flexibility.

7. Build, assemble, or purchase

 

Approach What It Is Best for Watch out
Build it yourself Integrate open-source components yourself, such as Backstage, Argo CD, and Crossplane Large organizations with a dedicated platform team A lot of maintenance; the expertise is concentrated among just a few people
Assemble Combining commercial building blocks, such as a SaaS portal, managed Kubernetes, and CI/CD as a service Organizations that want to get started quickly and remain in control License fees and multiple vendors
Purchase A platform-as-a-service offering, including management, such as Multistax from SUE Software teams without their own platform expertise Less freedom outside the platform

Whatever you choose, stick to one platform. According to State of Platform Engineering, Vol. 4 (January 2026), 55.9% of organizations run more than one Internal Developer Platform, resulting in duplicate management and differing standards.

Two examples: Tilaa built the Nexaa platform in collaboration with SUE; Stormin purchases the platform through Multistax.

8. The platform team

A platform team consists of platform engineers with expertise in Kubernetes, Infrastructure as Code, CI/CD, and security, as well as a product owner who determines what the platform delivers first. Start small: a team of a few engineers and a product owner can build the first golden paths. The book Team Topologies by Matthew Skelton and Manuel Pais describes the platform team as a team that serves development teams; those teams are the customers. For 24/7 management of the underlying clusters, you’ll need more people or a partner. See the Kubernetes guide, section 13.

9. Step-by-step guide

 

  1. Talk to developers and identify where software delivery is currently being slowed down—for example, the lead time from commit to production.
  2. Choose one common task that's particularly difficult, such as launching a new service or requesting an environment.
  3. Build the smallest platform that solves that task: the thinnest viable platform.
  4. Test it with one pilot team and make improvements based on their feedback.
  5. Measure the impact: adoption, turnaround time, and satisfaction.
  6. Expand to include more teams and more golden paths, and add a portal if there's enough to show.
  7. Treat the platform as a product: with a roadmap, releases, documentation, and support.

In the Platform Roadmap Workshop, you'll go through the first three steps in one week.

At Booking.com, deployment time for SAP environments went from days to less than an hour; at Tidalis, provisioning time went from hours to minutes.

10. Measuring success

According to State of Platform Engineering Vol. 4, 29.6% of platform teams do not measure the success of their platform. That’s why you should start measuring even before the first release. Without a baseline measurement, you won’t be able to demonstrate the platform’s value later on.

What Metric Source
Speed Time from commit to production, deployment frequency DORA Metrics
Stability Change failure rate, recovery time after an incident DORA Metrics
Adoption Percentage of services on a "golden path," active portal users The platform itself
Developer Experience Satisfaction, time until the first release by a new developer Survey and onboarding data
Costs Costs per team or service FinOps tools

11. AI and Platform Engineering

Nearly 90% of tech professionals use AI in their work, and 94% of organizations cite AI as a key factor in the future of platform engineering (State of Platform Engineering Report, Volume 4). The DORA Report 2025 describes AI as an amplifier: with a good platform, you deliver faster; with a messy platform, you make mistakes faster.

What this requires of the platform

  • AI assistants in the IDE and the pipeline will have the same safeguards as developers: scans, reviews, and policies.

  • AI agents that modify infrastructure or code operate through the platform, with their own identity, minimal privileges, and logging. Gartner cites identity and access management for AI agents as one of the security trends for 2026.

  • The platform provides models, GPUs, and vector databases on a self-service basis, so teams don't have to build their own AI stacks.

  • According to Gartner, no-code and vibe coding open up new attack surfaces. A "golden path" for AI-generated code keeps it within the same boundaries.

12. Why platforms fail

 

Cause How to Prevent It
Built without consulting developers Interviews up front and a pilot team
Too much at once Start with one golden path, then expand
Making it mandatory Making the path so appealing that teams choose it on their own
No product owner Someone who sets priorities based on what developers need
Multiple platforms side by side One platform with one set of standards
Not measuring A baseline measurement and fixed metrics from the start
Underestimating operations Capacity for upgrades, security, and support, or a managed services partner

To find out why developers avoid a particular platform, read our blog post on the adoption of an Internal Developer Platform.

13. Frequently asked questions about Platform Engineering

All frequently asked questions
DevOps is a culture and way of working in which teams build and run their own software. Platform engineering makes this feasible at scale: a platform team provides the tools and automation, so that not every team needs to have all the infrastructure knowledge itself.
An Internal Developer Platform (IDP) is the set of tools, services, and automation that development teams use via self-service to build, test, and deploy software. It often includes a developer portal, golden paths, CI/CD, Kubernetes, security policies, and observability.

Backstage is an open-source framework for developer portals, originally developed by Spotify and now a project of the Cloud Native Computing Foundation. It offers a service catalog, templates, and plugins. It is flexible, but requires in-house development resources to set up and maintain.

No, but it's the most commonly used foundation because Kubernetes standardizes deployment, scaling, and recovery. A platform can also run on managed container services or serverless infrastructure. Learn more in the Kubernetes guide.

Four metrics from the DORA study for measuring software delivery: deployment frequency, time from commit to production, the percentage of changes that lead to an incident, and recovery time after an incident. They show whether a platform makes delivery faster and more stable.

Adoption by developers, continuing to maintain the platform, and demonstrating its benefits. Having multiple platforms running side by side and a platform designed without input from developers are the most common causes of problems.

What should your platform provide first?

In a Platform Roadmap Workshop, you’ll work with SUE over the course of one week to determine what your platform needs to deliver first.

Plan a Platform Roadmap Workshop · Platform Engineering at SUE

Jonah Sanderse
Jonah Sanderse

Let’s talk


What should your platform provide first?

In a Platform Roadmap Workshop, you’ll work with SUE over the course of one week to determine what your platform needs to deliver first.

Plan a Platform Roadmap Workshop · Platform Engineering at SUE

* required

We will respond within one business day. By submitting this form, you confirm that you have read and understood our privacy policy.
SUE Cloud & AI Native Solutions
Privacy overview
This website uses cookies. We use cookies to ensure that our website and services function properly, to gain insight into the use of our website, and to improve our products and marketing. For more information, please read our privacy and cookie policy.