While artificial intelligence was supposed to drastically reduce the cost of software development, market realities and the rising price of AI tools have quickly verified these promises. In response to these budget challenges, business is increasingly turning to low-code platforms that offer financial stability and allow teams to build applications faster.
But let’s face it: the phase of unconditional admiration is now coming to an end. The market is entering a phase of maturity, and with it comes cool, mathematical verification. It turns out that the generative AI revolution has its hidden, extremely expensive obverse. Service valuations are rising, vendor business models are changing and companies are starting to ask themselves: is automated code generation really worth it to us?
At the same time, without too much publicity, another technology is coming to the fore that is redefining the cost structure in IT – low-code platforms. And it is the marriage of these two worlds that could prove to be the real breakthrough that business has been waiting for.
New supplier maths: When AI invoicing hits the ceiling
For a long time we lived in a market anomaly. The big language model providers offered their tools relatively cheaply, subsidising technology development in order to gain as much market share as possible. That phase is now behind us. Today, AI providers are starting to count their own costs – and these, given the gigantic demand for computing power and server infrastructure, are astronomical.
The effect? A market correction, which can most clearly be seen in the leaders’ price lists:
- Anthropic has clearly revised upwards the expected cost of using its advanced Claude Code tool.
- GitHub Copilot is steadily evolving towards usage-based billing models, moving away from simple flat-rate subscriptions.
- Budget reality: In extreme cases, in companies with very intensive usage patterns, single, advanced developers have been able to generate monthly costs for queries to the artificial intelligence API reaching six-figure sums.
Of course, at the level of the individual employee and small daily tasks, these costs seem marginal. The trap, however, lies in scale. When the demands of entire teams are added up, multiplied by months of complex projects, invoices land on managers’ desks that effectively cure uncritical AI-optimism. Return on investment (ROI) becomes a moving target, and previous budget assumptions start to resemble fortune-telling.
Low-Code as architect of the new financial order
In this landscape of uncertainty, traditional software development faces the same familiar walls. The IT job market – although more stable than in previous years – still suffers from a shortage of niche specialists. Deploying new developers takes time, and external agencies expect rates that immediately drain budgets with every additional project requirement.
And this is where the low-code approach enters the scene. It is not a new technology, but in the current economic reality it is gaining a whole new meaning. It allows the existing pyramid of competences to be inverted through the so-called democratisation of development.
Shifting the burden of developing simpler applications from the central, perpetually overloaded IT department directly to the business departments (marketing, sales, HR) dramatically changes the spending structure. Employees who do not have a classical programming background – so-called Citizen Developers – are able to build, test and modify work tools themselves.
Crucially from a CFO’s point of view: the maintenance and development of such applications is done at an incomparably lower process and computational cost. Instead of generating millions of lines of code with costly LLMs, which then have to be verified by a senior programmer anyway, the business benefits from proven, repeatable and low-maintenance visual components.
Smart blocks: The symbiosis of low-code and artificial intelligence
However, the biggest mistake would be to consider Low-Code and AI as competitors in the battle for budgets. The real revolution occurs where the two technologies meet. Modern Low-Code platforms do not reject AI – they domesticate it, offering AI functions as ready-made, integrated modules.
Instead of engaging a team of data engineers to build machine learning models from scratch, companies can use ‘smart blocks’. Some of the most popular scenarios include:
- Automated conversational systems: Chatbots and voicebots serving the customer, configured using visual editors.
- Predictive analytics and forecasting: modules capable of drawing conclusions from a company’s internal databases in seconds.
- Image and document recognition: Automatic processing of invoices, contracts or requests without manual transcription of data.
This gives small and medium-sized businesses tools that were previously reserved only for technology giants with unlimited budgets. Combining advanced AI functions with a company’s existing system architecture becomes simple and, most importantly, financially predictable.
New Kanban board: Evolution of roles, not downsizing
This paradigm shift naturally forces a new organisation of work. The vision in which artificial intelligence frees up people and completely replaces programmers has proven to be utopian. Instead, we are seeing a shift in emphasis.
| Feature | Traditional development (supported by pure AI) | Hybrid model (Low-Code + IT Architecture) |
| Main cost | High (AI computing power + Developer time) | Low/Medium (Platform licence + Coordination) |
| Implementation time | Medium (Requires testing, debugging of generated code) | Very short (Ready-made, tested components) |
| The role of the programmer | Writing and verifying code line by line | Architecture design, security, special logic |
| The role of business | Defining requirements and waiting for the result | Active creation and adaptation of tools |
In everyday business practice, standardised, repetitive activities are delegated to low-code platforms or automated. This gives skilled developers the space and time to do what really generates added value: solving complex logic problems, taking care of cyber security, optimising performance and designing global system architectures.
An excellent example is internal reporting systems, such as sales or management dashboards, which aggregate data in real time. In the past, the creation of such a tool required the appointment of a project committee, the writing of milestones over many months and the involvement of front-end and back-end developers. Today, thanks to the hybrid model, the sales department is able to launch a working dashboard in a few days.
Companies that approach this transformation strategically recognise that the biggest gain is not at all in mechanically slashing employee costs. The real currency is structural efficiency: drastically reduced time-to-market, smoother communication between business and IT and the flexibility to respond immediately to market changes.
Towards technological maturity
The software development landscape is no longer black and white. The future does not belong to the radicals – neither to those who want to write every line of code by hand like craftsmen, nor to those who trust unreservedly that generative algorithms will do all the work for free.
Pragmatism is winning. The future is a hybrid model that accurately weighs economic rationale. Where uniqueness, advanced logic and top performance matter, classical programming (supported judiciously by AI) will remain irreplaceable. However, where speed, adaptation to current business needs and control over budgets matter, Low-Code becomes the default choice.
