Software Defined Vehicles
Strategic liabilities
Why agentic AI exposes legacy systems
Agentic AI is widely regarded as the next evolutionary step in artificial intelligence. Yet according to a recent Cognizant study, most IT landscapes are not prepared. Technical debt, legacy systems and rigid budget structures are increasingly becoming strategic liabilities.
The German federal government has recently initiated legislation to implement the EU AI Act. With the Federal Network Agency designated as the central supervisory authority, it is clear that artificial intelligence in Europe is no longer viewed solely as an innovation driver but as a regulated technology domain governed by clear rules.
Companies will be required to demonstrate how their AI systems operate, how risks are assessed and who holds internal responsibility. The AI Act follows a risk-based approach, focusing particularly on high-risk applications. For the automotive industry, this is far from marginal. Advanced driver assistance systems, autonomous functions, AI-supported production control and even data-driven decision processes in sales may fall into regulated categories.
AI therefore becomes not only a matter of performance and scalability, but of governance, documentation and technical resilience.
AI scaling requires modern IT foundations
Advertisement
In this context, a recent Cognizant study entitled “AI in a Two-Year Window: The Path to Legacy Modernisation” gains strategic relevance. Based on an international survey of business and IT executives, the study argues that organisations seeking to scale AI within the next two years must fundamentally modernise their existing IT landscapes.
The urgency becomes particularly evident when examining agentic AI. These systems do not merely analyse data or generate recommendations. They plan tasks, coordinate multiple systems and execute end-to-end workflows within defined boundaries.
Agentic AI operates across system layers and assumes operational responsibility. In doing so, it directly exposes weaknesses in fragmented architectures, inconsistent data models and outdated governance structures.
A widening gap between ambition and readiness
According to the Cognizant study, 85 per cent of executives fear that their current technology base could hinder AI integration and scaling. At the same time, at least three quarters believe they can achieve their modernisation targets within two years.
This optimism contrasts sharply with structural reality. Only 17 per cent of respondents believe their existing infrastructure is capable of supporting agentic AI systems that coordinate multiple platforms and execute complex workflows autonomously. In other words, more than four out of five organisations see themselves as insufficiently prepared for the next stage of AI evolution.
Technical debt remains a central obstacle. Ninety-three per cent of organisations report having reduced no more than a quarter of their accumulated technical debt. Looking ahead to 2030, 79 per cent expect to have eliminated less than half.
For agentic AI, this is critical. Such systems depend on clearly defined interfaces, consistent data governance and stable security architectures. Without standardisation and clear accountability, the risks of flawed decisions, security gaps and non-transparent process chains increase substantially.
Advertisement
The legacy budget trap
The study also highlights a structural imbalance in IT spending. Currently, around 61 per cent of IT budgets are allocated to operating and maintaining legacy systems. By 2030, this figure is expected to drop to 27 per cent, with the released resources redirected towards innovation, automation and AI.
However, this financial reallocation will only materialise if modernisation efforts accelerate significantly. Otherwise, legacy maintenance will continue to absorb capital that is urgently needed for AI transformation.
Implications for the automotive industry
The automotive sector faces particular complexity. Software-defined vehicles, over-the-air updates, AI-driven production planning and data-based aftersales services already depend on deeply integrated digital infrastructures.
Agentic AI amplifies these dependencies. It interacts with ERP systems, MES environments, PLM platforms and fleet data streams. It orchestrates cross-functional workflows and operates within safety-critical and regulated contexts.
Under the EU AI Act, explainability, traceability and robust governance become mandatory, especially where AI systems influence vehicle safety or regulated production processes. Logging, monitoring and role-based accountability are no longer optional features but prerequisites for scalable deployment.
In this sense, agentic AI functions as an architectural stress test. It reveals whether organisations have truly modernised their IT foundations or merely layered innovation onto legacy structures.
Modernisation pace versus AI momentum
Cognizant’s conclusion is unequivocal: organisations must align the pace of IT modernisation with the speed of AI development. If companies succeed in approaching the two-year transformation window, they will be better positioned in an increasingly AI-driven economy.
If not, technical debt may evolve from an operational inconvenience into a structural competitive disadvantage. Agentic AI therefore represents more than a technological upgrade. It marks a turning point where incremental optimisation gives way to architectural redesign.
