No Code and AI: a threat or an opportunity for developers?

Over the last years, the software development industry has been revolutionised by two apparently disruptive tendencies: No Code/Low Code platforms and the usage of Artificial Intelligence (AI). Both promise a programming democratisation, making it accessible to a wider public and increasing efficiency over business processes. Are they really tools able to free developers from repetitive tasks and jargon, or do they hide pitfalls able to disrupt the quality, scalability and sustainability of software systems? This article analyses these technologies with a critical and concrete view.

The idea behind no-code/low-code platforms is easy yet powerful: many business processes follow recurrent schemes that are repeated in hundreds, if not thousands, of different applications. What's the point, then, in rewriting the same functionalities over and over again, like data input in a form, the recovery of information from SharePoint or their visualisation in a table?

The aim is to offer modular and intuitive tools that allow also those who don't have tech expertise to build useful and functioning applications, without writing a single code string. The promise is fascinating: more autonomy for business units, less IT dependency and a reduction of costs. But in reality, things are much more complicated.

Low-code and no-code platforms tend to generate high expectations, but they rarely are able to keep them over time. Business departments often discover they can't fully substitute the support of developers and the IT department. Sometimes, the attempt to 'simplify' translates into a paradox: solutions created to facilitate processes end up complicating them in the long run.

One of the main reasons is that programming means dominating a language – artificial, yes, but still complex. Like with a foreign language, vocabulary, grammar, practice and years of experience are needed to reach command. Those proposing no-code suggest replacing this knowledge with a deck of 'logical cards' to be combined: while it works to build easy sentences, it does not allow for the expression of articulated thoughts. The result? An application that works only until it remains within certain system limits.

Furthermore, these platforms mask the complexity but don't eliminate it. When technical problems arise – inevitable in evolving systems – they become hard obstacles to overcome. The promised abstraction becomes a cage.

Another issue is the technological dependence (vendor lock-in). Many low-code/no-code tools are proprietary, and moving from one to another often implies that everything has to be rewritten from the start. This is a cost and risk that not all companies can afford. Moreover, there is also a skill problem: if business departments do not fully know their processes, they risk creating inefficient or even detrimental solutions, exacerbating the weight on the IT department instead of lightening it.

However, not everything is to be discarded. No-code/low-code platforms can be useful for simple and low-risk tasks: creating a form, sending an email, and compiling a report. They can also be helpful in facilitating communication between developers and final users, for example, via interactive prototypes

To sum up, these platforms work well in low-complexity and high-predictability contexts, but they are not suited for the following cases: critical-mission systems, high availability, scalability or high requirements in terms of security, performance and maintainability. 

Artificial intelligence is deeply transforming developers' work too. Large Language Models (LLMs) like GPT-4 by OpenAI, Gemini by Google or LLaMA by Meta are already daily used by an increasing part of the tech community. According to Stack Overflow's Survey of 2024, 62% of developers regularly use AI, and another 14% foresee using it in the short term. Among the most widespread tools are ChatGPT, GitHub Copilot and Google.

The advantages are obvious: increased productivity, support in learning new languages, automatic error detection, automation of repetitive tasks. However, there are also risks that should not be underestimated: lowered code quality, copyright or licence violations, security vulnerabilities and potentially misleading answers given by models trained to always provide an answer, even when there is no solid basis.

The advantages are obvious: increased productivity, support in learning new languages, automatic error detection, automation of repetitive tasks. However, there are also risks that should not be underestimated: lowering the quality of code, copyright or licence violations, security vulnerabilities and potentially misleading answers given by models trained to always provide an answer, even when there is no solid basis. 

No Code, Low Code and Artificial Intelligence do not represent a direct threat to developers, but tools that must be understood and used with awareness. They can improve efficiency, facilitate communication and lower barriers to entry. However, they do not eliminate the inherent complexity of software development, nor do they replace the technical, analytical and design skills that underpin any well-built system.

For developers, the future will not be that of an obsolete profession, but of an increasingly strategic role: interpreters between technology and business, architects of robust systems, aware of the potential and limitations of new tools. The real challenge is not to choose between human and machine, but to integrate both intelligently.