Software and IT Industry in the AI Era

The way we write software has changed quite a lot in the last couple of years. Earlier, building even a small feature meant spending hours writing code, going through documentation, fixing errors and trying different approaches. Now, with tools like Claude, ChatGPT, Cursor and other AI coding agents, a lot of that work can be done in a fraction of the time.

And this is just the beginning. AI is not only helping us write code anymore. It is changing how we approach software development itself.

Prompting Is Becoming a New Programming Skill

We have spent years learning programming languages like JavaScript, Python, Java and many others. But now, there is another skill developers are spending more and more time learning: how to communicate with AI.

A simple prompt can generate a function or a component. But things get much more interesting when we start using advanced approaches like RAG pipelines, AI agents, agent networks, tool calling and context engineering.

For example, instead of manually building every part of a workflow, we can now create an AI agent that understands a task, accesses relevant data, uses external tools, executes actions and checks its own results. Multiple agents can even work together, with each agent handling a different responsibility.

This is quite different from simply asking ChatGPT to write a piece of code. We are slowly moving towards a way of programming where we describe what needs to happen, provide the necessary context and let AI handle much of the implementation.

Of course, knowing how to prompt is not enough. We still need to understand the underlying technologies, review the generated code and know when the AI is doing something wrong. But the way we interact with computers to build software is definitely changing.

Engineers Were Never Just Code Writers

There is something I think people often misunderstand when discussing AI replacing developers. Writing code was never the entire job of a software engineer.

Before writing a single line of code, an engineer needs to understand the problem, discuss requirements, think about the system design, choose the right technologies and consider what could go wrong. Even after development, there is testing, deployment, security, performance and maintenance.

Writing code is just one part of all this.

Think about how much of a developer's daily work actually involves repetitive things. Creating similar API endpoints, writing boilerplate, setting up components, fixing small syntax errors or implementing features that follow an already established pattern. These tasks take time, but they don't always require much engineering thinking.

This is exactly where AI is making a difference. A lot of that repetitive coding work is becoming automated. And honestly, I don't think this is necessarily a bad thing.

Why should an engineer spend hours writing something that an AI can generate in a few minutes, especially when the engineer already knows what needs to be done?

The important part is that someone still needs to understand the generated code, make sure it fits the existing system, handle unexpected situations and take responsibility when something breaks.

In that sense, AI is not necessarily replacing engineering. It is changing the work engineers spend their time doing.