The Rise of AI Coding Agents: Impact on Platform Engineering Teams
How AI coding agents like GitHub Copilot Workspace and Cursor are reshaping platform engineering. What teams need to prepare for and how to leverage these tools.
Spotify’s Backstage became the default Internal Developer Portal (IDP) because it solved the real problem: developers can’t find anything. Where’s the API docs? Who owns that service? How do I deploy to staging?
In 2026, Backstage is table stakes. The differentiator is what you build on top of it — and increasingly, that’s AI.
Instead of browsing a catalog, developers ask questions:
// Backstage AI Search Plugin
app.use(
createPlugin({
id: 'ai-search',
routes: {
root: aiSearchPage,
},
apis: [
createApiFactory({
api: aiSearchApiRef,
deps: { catalogApi: catalogApiRef },
factory: ({ catalogApi }) => new AISearchApi({
catalogApi,
llmEndpoint: 'http://ai-service:8080/search',
}),
}),
],
})
);Developer asks: “Which service handles payment webhooks?” — AI searches the catalog, README files, and API specs to give a direct answer with links.
The biggest IDP problem: stale docs. AI generates and updates them:
async def generate_service_docs(service):
"""Generate docs from source code, API specs, and infra config."""
context = {
'readme': await fetch_readme(service.repo),
'openapi': await fetch_openapi_spec(service.api_url),
'terraform': await fetch_infra_config(service.tf_module),
'dependencies': await fetch_deps(service.repo),
}
docs = await llm.generate(
system="Generate concise service documentation. Include: purpose, API endpoints, dependencies, deployment, and runbook.",
user=f"Service: {service.name}\n\n{json.dumps(context)}"
)
await backstage.update_techdocs(service.name, docs)Developers describe what they need in natural language:
Developer: "I need a PostgreSQL database for my new service,
production-grade, EU region, 100GB"
AI Assistant: "I'll create a Terraform module for you:
- Cloud SQL PostgreSQL 16
- europe-west4 region
- 100GB SSD, automated backups
- Private VPC connection
Here's the PR: [link]
Estimated cost: €180/month"This bridges the gap between “I need infrastructure” and “I need to learn Terraform.” The platform team maintains the golden paths (more on that in my Terraform Pilot content), and AI makes them accessible.
Software templates are Backstage’s killer feature. Define opinionated paths for common tasks:
# template: create-microservice
apiVersion: scaffolder.backstage.io/v1beta3
kind: Template
metadata:
name: create-microservice
title: Create a Microservice
spec:
type: service
parameters:
- title: Service Details
properties:
name:
type: string
language:
type: string
enum: [go, python, typescript]
needsDatabase:
type: boolean
steps:
- id: scaffold
name: Generate from template
action: fetch:template
input:
url: ./skeleton
values:
name: ${{ parameters.name }}
- id: create-repo
name: Create GitLab repository
action: publish:gitlab
input:
repoUrl: gitlab.com?owner=platform&repo=${{ parameters.name }}
- id: register
name: Register in catalog
action: catalog:registerI cover the Kubernetes side of this — the namespaces, RBAC, and resource quotas that back these templates — at Kubernetes Recipes.
Month 1: Deploy Backstage, import service catalog from GitLab/GitHub Month 2: Add TechDocs, integrate CI/CD status Month 3: Build 3-5 software templates for common tasks Month 4: Add AI search and documentation generation Month 5: Self-service infrastructure via AI + Terraform templates Month 6: Measure developer satisfaction and iterate
The automation foundation for deploying and maintaining Backstage itself? Ansible. I manage Backstage infrastructure-as-code using the patterns at Ansible Pilot — PostgreSQL setup, Redis cache, plugin configuration, all automated.
An IDP doesn’t eliminate the platform team — it amplifies them. Instead of being a ticket queue (“please create me a database”), they become product owners building self-service experiences.
That’s platform engineering done right. Not more tools — better abstractions.
AI & Cloud Advisor with 18+ years experience. Author of 8 technical books, creator of Ansible Pilot, and instructor at CopyPasteLearn Academy. Speaker at KubeCon EU & Red Hat Summit 2026.
How AI coding agents like GitHub Copilot Workspace and Cursor are reshaping platform engineering. What teams need to prepare for and how to leverage these tools.
Schedule Kubernetes workloads when and where the grid is greenest. How carbon-aware scheduling works, the tools available, and the business case for sustainable compute.
Deploy confidential containers on Kubernetes using AMD SEV-SNP and Intel TDX. Protect sensitive AI workloads with hardware-level encryption in untrusted environments.