Skip to content
  • AI
  • Development
  • Work

The new developer skill set

December 8 - 23

Rémi Prévost
Partner, Director ⏤ Software Development

With the birth of generative AI and incredibly powerful models like GPT, a whole new world of possibilities for software development has emerged. Artificial intelligence is revolutionizing the way we write code and ship software.

Coding assistants like GitHub Copilot and Tabnine can take a few lines of code and instantly provide relevant suggestions to complete them. They can write unit tests to cover multiple cases for complex features.

GitHub Copilot

Start a conversation about your codebase.

The barrier to entry for writing code is getting lower as tools like Framer (which allows any user to build a website from a prompt) and Phind (which literally writes code for the user) are taking their place in the market.

Other tools that can help developers achieve more with less are increasingly popular, such as What The Diff which improves code reviews by analyzing code diffs to provide useful insights. It’s only a matter of time before code review itself is truly taken over by AI, as more and more automated tools can spot code styling improvements, fix typing mistakes, etc.

What The Diff

Instant code refactoring.

If AI can write, explain, and find bugs in code, are developers still relevant? We absolutely think they are, but the “classic” developer skill set will need to be revisited to make the most of the tooling that’s now available to developers and focus on the unique value they can provide.

How will developers’ skills need to adapt?

These new tools all have the same clear goal: improve developer productivity. And because they integrate seamlessly with the tools developers already use, it would be a mistake not to leverage them. 

AI assistants have been trained on existing code—often open-source—to be experts at predicting code that developers want to write. If engineering teams want to stay competitive, not using an AI assistant today would be akin to using a hammer instead of a nail gun to build a house.

But if everyone is now using the same nail gun (i.e., AI), what are the distinctive skills developers need to possess to make sure they can build the best houses (i.e., digital products)?

Focusing on innovation & custom development

AI models are trained on existing codebases and products. This means they can’t be aware for the moment of things that haven’t been done yet, because they can’t have been trained on it.

For developers, it means that skills that involve breaking new ground, like figuring out how to connect to different devices together or integrating your codebase to a custom third-party provider, will remain highly valuable.

Current AI models have no logical thinking, hence cannot know for sure that something is correct, or cannot decipher complex problems. The ability to take on a problem—and not just requirements—  that needs to be solved creatively because it hasn’t been solved by anyone else will always be highly valuable.

Knowing the difference between what and how

At Mirego, we know that stuff like code maintainability, user experience details, performance fine-tuning, pixel-perfect alignment and copywriting tweaks are what makes a good digital product great. Knowing what to do—as well as knowing what not to do—are essential tools for the success of a digital product.

While developers can certainly leverage AI tooling to assist them with how to make these elements happen, the best ones are going to be the ones who truly know what to make happen. And on top of that, they will apply a level of care, love, and attention to detail which no AI will be able to match.

Keeping up with emerging technologies & practices

In order to be relevant, developers need to stay up to date on new technological advancement. For example, in the last 10 years, we have seen the rise of:

  • New functional programming languages
  • Decentralized computing, immutable data structures and blockchains
  • Reactive programming and declarative UI
  • Multiplatform development that targets an ever-increasing number of platforms
  • DevOps, DevSecOps, etc.

The days of developers learning a single language and using it for their entire career is likely over. They will need to constantly learn new programming languages, techniques, and paradigms. However, AI tools will be able to remove a huge amount of effort while doing so—it’s much quicker to learn a new language when there’s an assistant suggesting what you might want to write next instead of trial and error or copy/pasting answers from Stack Overflow.

Now that AI has taken over mundane tasks from developer workloads, it’s more important than ever to focus on deeper and cerebral tasks that can then be used in combination with powerful AI tooling.

Developers will continue to provide value

AI can help developers generate code, but not custom digital products. Developers that will be thriving in the future will be the ones who leverage AI tools while focusing their efforts on things AI can’t do (yet). They will let tools solve the how while they will focus on the what.

We believe it’s best to focus our development efforts on where we can actually provide the most value and let generic stuff be handled elsewhere (frameworks, boilerplate projects, tooling, etc.). This won’t change in the future—the tooling is just going to get better.


En français SVP !