Low Code, No Code, Vibe Coding: Potentials and limits of current software trends

Drag-and-drop instead of code? LCNC platforms (Low Code / No Code) and new trends such as vibe coding promise to drastically simplify the development of digital solutions. Providers such as Microsoft, SAP and newcomers such as Anthropic are developing tools that enable specialist departments and individuals to build applications themselves – at least that's the promise – almost entirely on the basis of natural language.

What sounds like democratization also raises critical questions: How sustainable are such solutions and are these promises being kept? Where are the opportunities – and where are the risks? The question isn't whether vibe coding or LCNC make sense – but where their useful boundaries lie.

No-code platforms are aimed at users without any programming knowledge. The user interfaces are highly visual so that applications can be assembled using drag-and-drop. This makes no-code ideal for simple applications that require quick adjustments, such as internal data collection tools.

Low-code platforms allow a little more flexibility and offer the option of adding code. This allows more complex applications to be created that are adapted to specific company requirements. These tools are aimed at users with basic programming skills, but are more accessible than conventional development platforms.

Vibe Coding goes one step further: here, AI not only writes individual lines of code, but entire components based on short descriptions in natural language. The vision: anyone who can formulate an idea should be able to build software – without any prior technical knowledge. The AI tool makes suggestions based on a language specification, the user accepts the suggestions and builds an application.

In practice, AI takes over many tedious tasks and increases productivity: it generates code, suggests functions, integrates libraries and can write tests. Research work is significantly reduced, and you can have things explained to you. Tools like GitHub Copilot or Amazon CodeWhisperer show how far this has already come.

But while some tech giants are already talking about the end of programming, we take a critical view of this evolution. Neither no code, low code, nor vibe coding can currently replace solid software development – at least not when quality, scalability, security, or maintainability matter. In our experience, the principle of “story in, code out” doesn’t work. Without clean architecture, documented interfaces, well-thought-out DevOps processes, and sufficient test coverage, the results remain mere prototypes – and quickly lead to problems.

Various well-known technology companies have specialized in the development and provision of low-code and no-code platforms and are focusing on the trend towards simplified software development. Platforms such as Airtable and Zapier, for example, offer no-code solutions that help users without programming skills to automate workflows and manage data. These tools are particularly attractive for companies looking for uncomplicated, easily configurable solutions.

In the low-code area, established providers such as Salesforce, OutSystems and Mendix offer platforms that allow a higher degree of flexibility and can also map complex enterprise applications. SAP Build and Microsoft Power Apps are also examples of low-code solutions that are increasingly proving their worth in the areas of process automation and data analysis. While no-code tools are particularly suitable for less technical users from specialist departments, low-code platforms offer developers more options for customization and integration, which enables companies to make a differentiated selection.

The first platform initiatives are also emerging in the area of vibe coding: According to media reports, Apple is working with Anthropic on a software solution designed to strengthen and improve vibe coding. The trend shows that AI-supported software development is becoming increasingly important. However, it’s equally important to assess the limits of its application realistically.

LCNC platforms offer a wide range of advantages – especially in scenarios where digital solutions need to be created quickly, pragmatically and with minimal involvement from the IT department.

1. Rapid implementation and agility
   Visual editors and ready-made components allow ideas to be transformed into functional applications in a short space of time. Specialist departments can react directly to new requirements, iterate and adapt solutions independently. This shortens the time-to-market and strengthens the teams' personal responsibility.
2. IT relief and process automation
   Standard applications and recurring tasks such as approval processes, data collection or status queries can be implemented without IT support. This frees up resources in the IT department, which can concentrate on more strategically important issues. At the same time, automation increases efficiency and minimizes sources of error.
3. Pragmatic solutions
   LCNC tools offer a practical and cost-effective alternative to traditional development, particularly in areas with stable requirements and manageable integration needs. They are ideal for prototyping, temporary tools or the digitalization of manual processes.
4. Important note on cost-effectiveness
   LCNC projects are not automatically cheaper. License costs, potentially high adaptation costs and limited expandability can lead to considerable follow-up costs in the long term. There is also a risk of vendor lock-in if companies become too closely tied to proprietary platforms. LCNC initiatives should therefore be evaluated at an early stage with a view to their entire life cycle – not just with regard to the initial development.

• Shadow IT
  If applications are created without the involvement of IT, uncontrolled isolated solutions (shadow IT) can emerge that are difficult to maintain and harbor security risks.
• Scalability and maintainability
  Many citizen developers build applications that are not designed for growth or long-term use. If key individuals leave the company, there’s a risk of losing critical knowledge. In addition, documentation and adherence to standards are often lacking.
• No test automation
  Many low-code or no-code applications lack a dedicated test environment. Changes are made directly in the productive application – without staging or clones. Automated tests are hardly possible in such environments as there is no stable test basis and central components for testability are missing. As a result, work is carried out directly on the live system – with a correspondingly high risk of errors during operation.

• Compliance and data protection
  In-house developments do not automatically meet regulatory requirements. Caution is required, especially when it comes to personal data.
• Vendor lock-in
  Dependence on individual platform providers can prove to be a strategic risk in the long term – especially if proprietary technologies or data formats are involved.

As soon as software is deeply integrated into core processes or has a long-term and business-critical character, modular systems are no longer sufficient. The requirements in terms of architecture, security, performance and expandability exceed the capabilities of LCNC platforms.

In such cases, you need:

• Experienced developers,
• well thought-out, structured software architectures,
• scalable infrastructures and
• DevOps-oriented development processes.

Even vibe coding cannot currently meet these requirements. The speed and simplicity of code generation can be misleading – building robust software involves much more than what’s visible in the UI. Architecture, testing, security and operation cannot be "prompted away". Vibe coding is a helpful addition – but not a replacement for experienced development teams. Vibe coding quickly reaches its limits, especially with strategically important software: While LLMs deliver impressive results for prototypes, they often do not understand what they are actually generating. The stochastic nature of the generation means that the output is no longer consistent or suitable at some point – especially if the context becomes too large or key details are missing. Reverse engineering or targeted further development ("debugging") is hardly possible.

Before starting a project with an LCNC platform or based on Vibe Coding, the following questions can help:

• Does the tool need to communicate with multiple systems?
• Are data protection and security requirements relevant?
• Will the solution be operated and further developed in the long term?
• Is a large number of users and a high volume of data to be expected at some point?

If the answer is yes: It’s better to rely on traditional software development (preferably with the support of AI assistants) in order to obtain a high-quality, scalable, secure and low-maintenance solution.

Low-code and no-code platforms as well as vibe coding open up new ways of developing applications quickly and can relieve the burden on companies and professional development teams in certain areas. In some cases, they offer specialist departments the opportunity to act independently and digitize smaller processes more efficiently.

Nevertheless, they are no substitute for professional software development. As soon as complexity, integration or strategic importance come into play, they are only suitable to a limited extent. What matters most is a realistic assessment of the requirements – with the entire application lifecycle in mind.