Don't Panic — Internal Developer Platforms

An Internal Developer Platform is the product that lets developers ask for delivery capabilities without personally assembling every source-control setting, pipeline, cloud resource, and permission. It is not a portal with a confident color scheme. The portal is merely the bit that can be blamed in a screenshot. The platform is the contract and the machinery behind it.

The useful picture has three parts. The experience layer helps people find a capability, say what they need, and see what happened. The control layer validates the request, applies policy, coordinates providers, and records state. The resource layer is where source control, CI, clusters, identity, data, and observability systems do their specialized jobs. Mixing all three into one enormous workflow is how a helpful shortcut becomes a small weather system.

Start with a capability, not a tool. “Create a production-ready service” is a capability. A Git provider, pipeline, runtime, and telemetry stack are providers used to fulfill it. The platform should promise inputs, outputs, defaults, constraints, state changes, support, and lifecycle. That promise is what lets a user request the outcome without needing to become an archaeologist of internal tooling.

The surprise is that creation is only the opening scene. Generating a repository is a task with a completion result. A database or deployment configuration is a managed resource with desired state, updates, drift, backup, and deletion to consider. A green workflow can therefore be correct and still be nowhere near finished. The next tab explains the contract techniques that make this distinction usable.

A software catalog supplies the shared map: systems, components, APIs, resources, owners, and relationships. Templates can begin repeatable journeys, but neither catalog entries nor templates maintain themselves. The Intro explains the architecture and operational boundaries. Slides compress the relationships into a map. Cheatsheet turns them into review questions. Field Notes covers the expensive mistakes that appear once the platform meets real provider failures.

Keep security and reliability in the picture from the beginning. Self-service means authenticating callers, authorizing their scope, using narrow provider identities, retaining an audit trail, and exposing status when dependencies fail. Begin with one vertical slice that is discoverable, governed, fulfilled, observable, and supported. Diagrams are polite; real journeys are less easily impressed.