Low-code is dead: how the world of IT development has changed

We defended low-code platforms for a long time: we wrote articles, addressed objections and misconceptions. The reason was simple: low-code used to help businesses because it simplified and stabilized development through ready-made templates and built-in integrations.

However, years later it became clear that the concept has problems: - the number of integrations keeps growing, which means more time has to be spent creating and reworking them; - junior developers need extensive training so they can understand existing integrations and build something on their own without mistakes; - senior developers burn out from repetitive tasks - simply "maintaining" old integrations is a pretty monotonous and uninteresting job.

So integrations have grown, and there are no fewer routine tasks; simply clicking something together on a low-code platform is no longer the solution. Then AI appeared in development. Here is what made its use attractive to IT specialists.

How a developer solves a task: opens StackOverflow and GitHub, writes code, edits it, tests it, and only then deploys it. This usually takes hours or even days. AI performs the same tasks in seconds: on request it analyzes thousands of repositories, makes corrections, and delivers a finished result. GitHub even ran an experiment: it split developers into two groups, asked them to write an HTTP server in JavaScript, and timed them. One group used AI, while the other did everything on its own.

The result: developers assisted by AI completed the task 55% faster.

The DORA.dev team ran a survey and found that more than 75% of developers are more productive when they use AI assistance. The GitHub team got almost identical results. In their study, most developers reported feeling more productive, getting distracted less, and solving tasks faster.