🎧 This article is based on episode #127 of our podcast. You can listen to it here.
AI coding: an overview
AI coding has gained enormously in importance in recent months. Tools such as Lovable, Bolt, Replit, Cursor, Windsurf and others have become indispensable for many developers and non-developers. A distinction must be made between tools such as Lovable and Bolt, which build prompt-based frontends, and Cursor or Windsurf, which act as AI assistants in coding consoles.
But they all have one thing in common: the barrier to entry into software development has been lowered - and that's right! But before you start "vibe coding" your app, you should definitely be aware of certain risks.
Advantages and risks of AI coding
The main advantage of AI coding is the ability to create complex applications without in-depth programming knowledge. In Lovable and Bolt, you can simply explain to the AI via a chat window what kind of app you want to build. This opens the door to an even wider target group than no- and low-code already did. One major difference, however, is that with no-code tools such as FlutterFlow, you have to spend a few days or weeks familiarizing yourself with the tool. With Lovable, you only need to be able to write. A usable app can be ready in just a few minutes.
These tools are particularly suitable for building prototypes or entire websites in just a few minutes. Alex emphasizes that these use cases are on the safe side when it comes to data protection.
Because the accessibility of these tools also comes with risks. Many no-coders simply don't know what professional developers have learned about security and data protection over the years. And so API keys or even personal data can suddenly be accessible to unauthorized persons.
Another disadvantage is that, as a layman, you don't necessarily know how to proceed with the AI-generated code. Lovable makes it super easy to connect Supabase as a backend, but what if you want to move with the frontend? And how do you know if the code that was generated was really written cleanly? In some cases, code can even be protected by copyright. So if the AI simply copies an open source code that falls under a copyleft license, this can even have legal consequences in the worst case.
Security and data protection
An important topic in connection with AI coding is data protection. And not whether your user data is safe with you, but what the AI does with this data. This is because many of these tools use the data provided by users for training purposes. Alex points out that you should be careful not to share any sensitive information. Some providers offer the option to opt out of data usage, but this is often not the case by default.
The future of AI coding
The future of AI coding looks promising. The integration of AI into the development process could mean that traditional coding skills are less in demand, while the ability to work effectively with AI tools becomes more important. This could also change the way companies approach their development projects by integrating AI-powered tools more into their processes.
No-code tools such as WeWeb, Bubble and Xano already have their own AI assistants, which are getting better and better over time. And WeWeb makes it possible, for example, to import code from GitHub and generate an app from it. This means that the time-to-prototype or time-to-market is getting shorter and shorter.
But despite all the AI assistants in no-code tools, Alex would tend to learn at least the basics of programming with AI assistants. You need these basics and the way of thinking as a developer in both coding and no-code anyway.
Conclusion
AI coding offers enormous opportunities to revolutionize software development by facilitating access to complex development processes. At the same time, developers and companies need to keep an eye on the associated security and data protection risks. The future will show how these technologies will evolve and how they will change the role of developers and the structure of development projects.
