For more than two decades, enterprise software has been shaped by a logic of standardisation at scale.

Large SaaS platforms have become the backbone of digital business and public sector systems alike. They have enabled faster implementation and access to mature capabilities without building from scratch. In many cases, they have played a central role in accelerating digital transformation.

However, the experience of recent years points to a more complex reality.

Across both enterprise and public sector organisations, platform adoption has introduced dependencies and cost structures that are increasingly difficult to rebalance. The rise of MACH and hybrid architectures reflects this shift. In manycases, these approaches have emerged not primarily from architectural ambition, but as responses to the economic and operational constraints of large platforms.

At the same time, cost dynamics have evolved. Software and cloud now account for a growing share of IT spending, and organisations consistently report challenges in managing SaaS consumption and cost predictability¹. The introductionof AI-based features - often priced on a usage basis - has further increased variability in operational expenditure².

When operational cost begins to shape strategy

As platforms become deeply embedded, their economic logic begins to influence strategic decision-making.

Licensing models, consumption pricing, specialised skills, and integration overhead gradually shift spending toward OPEX. This reduces the capacity to invest in new capabilities. In practice, organisations often find that developmentbudgets are increasingly absorbed by maintaining and extending existing platform landscapes.

This has structural consequences. When core capabilities are tightly coupled to platforms, the ability to evolve independently becomes constrained. Strategic direction becomes partially dependent on vendor roadmaps and commercialmodels.

A changing foundation for building software

At the same time, the foundations of software development are changing.

AI-driven software engineering affects the entire lifecycle of delivery. It shortens the path from idea to implementation, enables earlier validation, and introduces new levels of automation in testing and maintenance.

For organisations already operating with agile methods, this strengthens existing practices rather than replacing them. Iteration cycles become shorter, and decisions can be made with better empirical grounding.

Empirical evidence supports this shift. A controlled study by GitHub found that developers using AI-assisted tools completed tasks up to 55% faster³. Similarly, McKinsey & Company reports substantial productivity gains across multiple stages of the software development lifecycle⁴.

As a result, the economics of custom software are changing. Building and evolving tailored solutions is becoming both faster and more manageable than before.

Reconsidering the role of platforms

In this context, organisations are reassessing the role platforms should play.

Platforms remain highly effective where standardisation, scale, and mature functionality are required. However, their limitations become more visible in domains with rapidly evolving requirements.

Cybersecurity is a clear example. Threat landscapes, compliance requirements, and response mechanisms evolve continuously. In such environments, the ability to adapt quickly can outweigh the benefits of adhering to predefinedplatform capabilities.

The same applies to areas closely tied to competitive differentiation or evolving business logic. This leads to a more granular architectural approach, where platforms and custom-built capabilities coexist based on where they create the most value.

What we are seeing in practice

From our perspective, this shift is already visible across organisations.

Platform usage is becoming more selective. Capabilities are moved out when costs or constraints outweigh their value. At the same time, custom layers are increasingly built around existing platforms to handle critical logic, integrations, and user-facing functionality.

In new initiatives, custom software is more frequently considered a viable option from the outset, particularly in areas where flexibility and speed are essential.

We also observe a shift in how maintenance is perceived. AI-driven engineering changes how systems can be evolved over time, reducing the historical burden associated with custom solutions.

These observations are not theoretical. They emerge directly from implementation projects, operational responsibilities, and long-term partnerships with organisations navigating complex digital landscapes.

From integrated procurement to modular ecosystems

This shift is also reflected in public procurement structures across Europe.

EU procurement is governed by harmonised directives and structured through classification systems such as the Common Procurement Vocabulary (CPV), which standardises how contracts are defined and categorised across all memberstates⁵. Procurement itself is typically divided into services, supplies, and works, each governed by specific rules and thresholds⁶.

In practice, this creates a modular procurement model.

Instead of acquiring integrated, end-to-end solutions, organisations increasingly procure:

• software (often as SaaS or packaged solutions)

• consulting and advisory services

• development capacity

• operational and infrastructure services

often through separate contracts and procedures.

This fragmentation is reinforced by procurement mechanisms such as framework agreements, dynamic purchasing systems, and competitive procedures that emphasise comparability and standardisation⁶.

The effect is a structural shift.

Responsibility for architectural coherence, integration, and long-term evolution is moving from suppliers toward the contracting organisation itself.

And it is precisely in this space that new development models such as AI-driven software engineering become relevant.

A broader perspective: dependency and autonomy

This development also has a strategic dimension.

European organisations operate in a landscape dominated by a small number of global technology providers. This dependency spans infrastructure, platforms, and increasingly AI capabilities.

EU policy initiatives increasingly emphasise digital sovereignty and resilience, highlighting the need for greater control over critical digital infrastructure and services.

As the cost and complexity of building software decrease, organisations gain more realistic options to develop and maintain selected capabilities independently. This does not eliminate the role of platforms, but it creates a more balancedposition from which to decide where dependency is appropriate and where autonomy is preferable.

Custom as a new baseline

The current trajectory does not point to the disappearance of platforms. It points to a change in their role.

Custom software is no longer limited to edge cases where no suitable standard solution exists. It is increasingly becoming a viable starting point in areas where flexibility, speed, and control are critical.

For this shift to materialise, however, organisations must actively create space for it.

Within EU procurement frameworks, this means engaging early in the process. Procedures such as competitive dialogue and innovation partnerships explicitly exist to address complex or not-yet-defined solutions⁶. Yet in practice, manyprocurements are still structured around predefined categories and solution assumptions.

Without early market engagement and informed requirement-setting, emerging approaches are unlikely to be considered.

From our perspective as a system integrator, this is not an abstract discussion.

We work closely with established platform ecosystems while simultaneously investing in how AI-driven software engineering can be applied at scale, under real governance and operational constraints.

This shift is already underway.

Custom software will not replace platforms.

But it is increasingly redefining the baseline from which architectural decisions are made.

1. Flexera – State of the Cloud Report 2024
   

2. Gartner – Cloud and SaaS cost trends
   

3. GitHub – The Impact of AI on Developer Productivity
   

4. McKinsey & Company – Unleashing Developer Productivity with Generative AI
   

5. Common Procurement Vocabulary – EU classification system for tenders ()

6. European Union – Procurement directives and procedures ()