How AI and DevOps Are Changing Software Development
Joshua Was Right: How AI and DevOps Are Changing Software Development
WarGames (1983)
In 1987 a film was released in Italy that, for many people of my generation, worked as a doorway in: WarGames.
I was nine years old when I first saw it, and yesterday evening I watched it again (for my second time!). Joshua, the NORAD supercomputer, was to all intents and purposes an artificial intelligence that learned from its own mistakes, and we are talking about a screenplay written in the Seventies. That film captured me completely. The modem connection, that sound which a few years later would enter every home, felt like the most magical thing in the world. At nine years old, I decided that one day I would write software.
Forty years later, the AI that learns by itself is no longer science fiction. It lives inside our IDEs, our deployment pipelines, the systems that monitor applications while we sleep. And the question we ask ourselves today is no longer “will AI arrive in software development?” but “how are we working with it right now?”.
For anyone commissioning a digital project, whether a complex website, a platform, or a business application, understanding this shift matters. Not for technical reasons, but because it directly affects the value and the longevity of what you will receive.
Software Development is no longer just “Writing Code”
For years the work of a developer has been described in simple terms: someone writes lines of code, someone else puts them online. Today that description is seriously incomplete.
Modern development has become an ecosystem where three layers coexist: the code itself, the artificial intelligence that assists whoever writes it, and the automated infrastructure that carries that code from the developer’s computer all the way to end users. This last part has a name, DevOps, and over the last couple of years it has turned into something far more sophisticated than it was even back in 2023.
AI as a colleague, not as a replacement
When an experienced developer sits down to work today, they have tools like GitHub Copilot, Claude Code, or Cursor working alongside them. These are not toys: they are assistants that read the context of the project, propose implementations, catch bugs before they become problems, and in some cases write entire portions of code under human supervision.
For a client, this does not mean a project will cost less or take half the time. That would be a convenient but misleading simplification. It means something more interesting: with the same investment, today you can achieve more, and you can do things that used to be economically unsustainable.
Think about project documentation, historically the weak spot of every software project. With AI alongside, producing and keeping it up to date is no longer a luxury reserved for large budgets, but a normal practice. Think about code quality: AI catches errors that a tired human eye misses, and this translates into fewer bugs reaching production, fewer emergency calls, less time spent repairing and more time spent growing the product. Think about automated tests, security reviews, optimizations: activities that were once squeezed for budget reasons, and that today fit naturally into the work cycle.
In other words, the value is not in saving days of work. It is in using those days better, on what truly serves your business.
A word of caution, though. All of this works only if at the center there are still people who understand what you are asking for, who translate your business goals into sensible architectures, and who know how to say no when AI proposes a dangerous shortcut. The real risk of projects “made by AI” without genuine oversight is exactly this: code that works today and falls apart in six months, because no one took responsibility for truly understanding it. AI accelerates those who already know where they are going. Those who do not, with AI, simply get lost faster.
DevOps: the infrastructure that manages itself
The other silent revolution is DevOps. It is the invisible part that makes sure users do not notice when an application is updated. That when a server has a problem, the system routes around it on its own. That when traffic spikes suddenly, the platform holds up without anyone having to step in at three in the morning.
Today DevOps is increasingly “intelligent”: monitoring systems no longer just report problems after they happen, they predict them. Deployment pipelines test automatically, roll back if something goes wrong, and optimize the use of cloud resources so that the client does not waste money.
In 2026, choosing a technology partner no longer means choosing someone who “knows how to program”. It means choosing someone who can orchestrate three things well together: skilled developers, AI used with judgment, and a modern DevOps infrastructure.
When one of these three legs is missing, the project limps along. Maybe not right away, maybe a year from now, when you need to grow it or adapt it to new requirements, and you realise it was built without thinking about tomorrow.
Joshua, at the end of WarGames, learns on his own that the only winning move is not to play. Forty years later, we developers have learned something similar: the winning move is not to chase every new technology, but to understand which technologies to use together, and how, in order to build something that actually works.
At Commpla
At Commpla we live this craft every day, with developers working side by side with AI, DevOps pipelines running on their own, and projects designed to last and grow over time. If you are thinking about a digital project, a platform to modernize, or you simply want to understand how these themes could affect your business, get in touch.
Let’s talk about it over coffee, virtual or real. 🙂