The End of DevOps?

There’s a pretty good chance that DevOps jobs as we know them are going to disappear.

We’ve heard this all before of course - the “job losses due to automation” story isn’t new - but surely an entire section of the industry dedicated to the automated shipping of code is going to be one of the hardest hit?

If Infrastructure-as-Code (IaC) is text based/language based, just like code, and if LLMs are really good at automating language and code, it’s not a great leap to think that IaC and DevOps as a discipline will be fully automated and hence that you’ll no longer need a large team to manage your pipelines.

Here’s what we’re seeing in the real-world:

• Client One: Replaced their six-person DevOps team with an automated pipeline that handles deployments, monitoring, and first-line incident response. The team? Made redundant.

• Client Two: Implemented platform engineering with such sophisticated automation that their “DevOps engineers” now spend most of their time babysitting systems that rarely need human intervention. Half the team has already been moved to other projects, where they’ll automate the release pipelines elsewhere.

The irony here of course is that the DevOps teams built the very automation that’s replacing them. Every CI/CD pipeline, every infrastructure-as-code template, every monitoring dashboard was designed to reduce human intervention. They’ve succeeded brilliantly.

So what about us humans? Well we’re not at a critical mass of automation - yet. There’s still a stack of stuff running manually; processes that are hand-cranked, a labyrinth of tools that are configured with precision and human-to-machine handoffs that need supervised.

Then there’s all the important stuff - the solution architecture, the strategic thinking, the business context. The crisis management when the automation fails spectacularly (and it does). But that said, those skills don’t require a full DevOps team, one-for-one versus today’s roles. They require a smaller number of very good people who understand both the technology and the business.

If you’re in DevOps, don’t panic. But we all need to evolve, and fast. Focus on the things AI can’t do yet e.g. stakeholder management, architectural decisions, and crisis leadership. Because the days of manually configuring deployment pipelines are numbered.