There is a lot of doom and gloom about what AI will do to jobs and industries. Some of that concern is fair, but I think it is just as important to look at the opportunities ahead.
In our field, it helps to see this not as the great replacement, but as abstraction and power. I have learnt more in the past year than in any year of my life, because finding, absorbing and retaining knowledge has never been this easy or this fast. I have always liked learning across different domains and sciences, and now I have a patient teacher on call for any of them. So yes, some of my skills will fade — the exact syntax for declaring a property in a given programming language, say — but what I gain in return is worth far more.
This is the abstraction part, and it is nothing new. A traditional builder happily welcomed power tools over manual, labour-intensive ones. Programmers years ago welcomed the move from machine language, to low-level languages, to higher-level ones. Using models, agents and natural language to build is the next rung on that ladder, and the most exciting one yet. The goal has not changed: to solve real problems and build things that matter.
What has changed is how much one person can carry. Knowing the first principles, standards and patterns of something well is what turns knowledge into action quickly. Because I know infrastructure reasonably well, the likes of AWS, Azure, CI/CD and Terraform, I can lay the bones myself and then point AI at the rest, keeping watch as it builds. It is worth remembering how far this role has already moved. We went from developers building software and throwing it over the wall to an ops team to deploy on-prem, to developers doing it themselves under DevOps, to dedicated DevOps teams, and now it is compressing again, with dev and ops merging back into one. The same is happening to analysis, product and quality. The roles are being warped together, and someone with enough experience can now cover ground across many different roles.
That, for me, is the real trade-off. Take finance, banking and payments, where I have spent a good part of my career. I no longer have to pour all of my time into a single subdomain such as treasury or lending. I can still build in those areas, but I can also extend my reach across the whole board: the standards, the regulators, the legislation, the lowest level details, and how it all fits together end to end.
What would once have taken a team, or several teams, many months if not years, can now be done in a fraction of that. The cost of building is not zero, but it is far lower. Solo developers are shipping real products on their own. Writespace.io would have taken me many, many months to build not long ago, and I certainly would not have built it on the side.
Don’t get me wrong, the risks are higher too. The testing, the gates, the security, the infrastructure, the cost of running it all — a lot of this gets turned on its head and has to be rethought. So no, it is not all rainbows and butterflies. But these are the kinds of problems we are good at, and we will figure them out.
So if you build software, I think this is the best time there has ever been to do it, with one caveat. The work is steadily moving up a level: from writing the code to composing whole systems and deciding what is worth building in the first place. Lean into that and you have more leverage than any developer in history has had. The risk was never really that AI would replace you; it is staying head down in the syntax while the value moves up to judgment and composition.