My thoughts on coding with AI

We’re at a point where you can build an entire application literally by talking to an AI.

You send a voice message in your favorite app (Discord, Telegram, maybe WhatsApp or LINE, perhaps even Signal, and so on) and go for a walk. A few minutes later, you get a notification about a new deploy on your app, with a big, exciting new feature in your product. Users can barely keep up with so many updates. The bottleneck in software development has changed.

And what about the code from that deploy, made from a voice message? Code that follows every rule you set up as a developer for the project, follows the best practices of system design, like TDD, red/green/refactor flow, BDD, DDD, a well-built test pyramid, manual testing by AI with Playwright or similar, good code coverage, mutation tests, following the project’s design system, in line with the latest code documentation, good OOP practices applied, following performance guidelines, running benchmarks, running plenty of security tests and validations, and with everything working well, shipping a deploy.

It takes time for you or your team to have this setup in place, and it keeps evolving, but once everything is ready, we basically have your style (or the team’s standard) being replicated by AI absurdly faster than a human would.

Our main job has become planning together with the AI and reviewing the code it produced to make any needed adjustments, and those adjustments are used to evolve the AI so it doesn’t repeat the same mistakes.

In 1897 humanity invented the first screen, in 1939 what can be considered the first computer, programming was done by manually changing wires and switches and punching cards that were processed as binary data.

In 2026, we talk to AI with our voice to build entire applications for the most varied platforms out there, going through every layer of software engineering. You can build solutions quickly, even games.

It’s getting harder and harder to predict the future and how technology will evolve, but we can use it to help us do amazing things and solve problems that had no solution or were hard to solve.

While some people refuse to use AI in any layer of software engineering, I’ve never been so excited about how easy it has become to solve problems and to get ideas out of my head and ship them, thanks to this high productivity, since before it took a big chunk of my free time.

So my view is that we should keep studying and improving ourselves to make AI and our other tools even better, aiming to solve even more problems and create even more amazing things even faster.

Tomorrow there will surely be many new tools, features, and updates, like there always are, but even faster and even better thanks to using AI as a tool that empowers humans.

In a world with so many things created 100% by AI, this text was written by hand, so if you made it this far, thank you so much.