DevOps Isn’t Dead—It’s Getting Buried Under Its Own Platform

I’ve sat through my fair share of platform engineering sales pitches.
I saw polished diagrams of “golden paths,” abstractions on top of abstractions, and internal portals meant to hide all the messy infrastructure details from developers. And every time, I found myself thinking the same thing:
This all sounds great—until I imagine my engineers needing a certification just to configure, maintain and use the thing.
I get the intent. Internal developer platforms promise consistency, security, and scale. But in too many cases, they’ve created more friction than they’ve removed. They’ve taken the spirit of DevOps—empowerment, speed, and ownership—and replaced it with complexity, bottlenecks, and new gatekeepers.
I’m not convinced we need most of it.
What Happened to “Self-Serve, With Guardrails”?
Back when I was leading DevOps teams, our mantra was simple: make common tasks easy, and dangerous tasks hard. That meant giving developers self-serve tools they could trust, with sensible defaults, documentation, and automation in place. Not freedom without safety—but also not safety without autonomy.
It worked. CI/CD pipelines ran through GitHub Actions or Bitbucket Pipelines. Infrastructure was templatized with Terraform or CloudFormation. Rundeck let us wrap ad-hoc scripts in guardrails and role-based access. The system was flexible, low-cost, and understandable.
Compare that to many modern platforms today:
- Custom DSLs and YAML formats that no one outside the platform team understands
- “Golden paths” that are broken or poorly documented
- Onboarding flows so convoluted you need another team’s help just to push code
All in the name of “developer experience.”
Platforms Weren’t Supposed to Be a New Discipline
At some point, we lost the plot. Platform engineering went from being a lens on DevOps to being its own full-blown discipline. Now, you need a platform team to interpret the platform. You need governance for the governance. Meanwhile your developers? They’re filing tickets again.
In short: we didn’t solve the problem—we just moved the bottleneck.
You Don’t Need a Platform to Move Fast
Most teams don’t need a “platform.” They need:
- Automation for repetitive tasks
- Secure, reliable deployment paths
- Clear documentation and templates
- A way to experiment safely
You can get all of that with:
- GitHub Actions for CI/CD
- Terraform modules for infrastructure
- AWS-native services and templates
- Tools like Rundeck for low-friction & end user friendly automation
- And a culture of collaboration and ownership
No internal portal. No six-month roadmap. No reinvention of Git.
AI Is Coming—And Simplicity Wins
Here’s the irony: as we invest in complex internal platforms, the industry is shifting again.
AI copilots and autonomous agents are entering the scene—and they thrive on clarity, consistency, and documentation. They don't need a shiny portal. They need a well-labeled Git repo and a predictable set of workflows.
If your internal platform requires tribal knowledge or Slack messages to understand, it won’t just frustrate humans—it’ll confuse your AI assistants too. I'm certain these platform providers will bake in their own AI agents to help. However this will only drive you deeper into dependence on the particular platform, with an agent that doesn't understand anything about your systems outside of it.
Build Like It’s Your Product
If you do go the platform route, treat it like a product:
- Start with your users (developers)
- Validate with real workflows
- Ship incrementally
- Document like it’s public
Otherwise, skip the platform and invest in DevOps that works—automation with empathy, infrastructure with clarity, and tools that feel like tools, not puzzles.
DevOps isn’t dead. But we should stop burying it under layers of complexity and call it progress.
Phil Austin is SVP of Product Engineering at Plus Company, where he leads engineering for the AIOS division. He has over 20 years of experience in Software Engineering and DevOps and has spent the past two years focused on advancing the intersection of AI and software development. Follow him on LinkedIn for more insights.