Why traditional application management is reaching its limits

For a long time, companies have operated their application landscapes with horizontally oriented Application Management Services (AMS) models: standardized services that used the same processes and technologies across all industries. This approach was efficient – as long as the requirements remained manageable.

Today, the reality is different. Specialist departments are actively driving technological innovation, new business models are emerging on a weekly basis, and regulatory requirements – especially in industries such as banking, healthcare, and the public sector – are becoming increasingly complex. Horizontal AMS models can no longer comprehensively cover these domain-specific requirements. What is needed instead is a deep understanding of the industry, combined with technological excellence and end-to-end responsibility across the entire application lifecycle.

This is exactly where modern, vertically integrated AMS approaches come in – and GenAI becomes a key accelerator.

GenAI in Application Management Services: More than automation

When people talk about GenAI in AMS today, many initially think of chatbots or automated documentation. That's not enough. The real potential lies in the deep integration of intelligent AI agents into operational processes – from error detection to deployment automation.

A key example is incident management: Instead of employees manually searching through logs and identifying sources of errors, autonomous AI agents semantically analyze system data, determine the most likely cause of the error, and automatically trigger workflows to fix it. What used to take hours now happens in minutes without human intervention. The mean time to repair (MTTR), i.e., the average time it takes to fix a problem, is measurably reduced.

The principle of proactive self-healing goes one step further: modern platforms detect the first signs of bottlenecks, such as an unusual increase in memory consumption, and intervene before agreed service level agreements (SLAs) are even compromised. Resources are automatically scaled and services are restarted preventively. Outages become the exception, not the rule.

The situation is similar with deployment automation: AI-supported systems check code changes, generate configurations for container environments, and control deployment pipelines fully automatically – including the creation of technical release documentation. The result is faster release cycles and a noticeably shorter time-to-value, i.e., the time from the start of development to the productive use of new functions.

Finally, specialized agents also take over vulnerability management: They continuously monitor newly discovered security gaps, evaluate their relevance for the respective system landscape, and deliver prioritized recommendations for action – proactively, not reactively.

Governance and control: What CIOs really need

For CIOs, the question of control is central. GenAI platforms used in the AMS context must meet enterprise requirements: role-based access rights, complete audit logs, zero-trust principles, and the ability to process sensitive data either on-premises or in certified cloud environments.

Especially in regulated industries, whether banking supervision, the General Data Protection Regulation (GDPR), or healthcare requirements, this is not optional, but mandatory. Platforms that are modular in design and can flexibly integrate various AI models (e.g., from OpenAI, European providers, or customer-owned models) without having to adapt interfaces or processes offer a decisive advantage here: they prevent dependencies on individual technology providers—so-called vendor lock-ins—while enabling cost-efficient use of the rapid progress in the AI market.

For the CIO organization, this means full transparency regarding costs, performance, and utilization rates of all services through integrated dashboards, combined with a controllable AI layer that meets compliance requirements.

Success stories from practice: What already works today

Theory and practice often diverge, especially in the field of AI – which is why it is worth looking at real-world implementations. At BayernLB, for example, a GenAI-based internal assistant was introduced that securely processes sensitive bank documents in the company's own cloud, optimizes knowledge access, and meets the highest data protection requirements. The result: less search effort for employees, faster decision-making processes, demonstrable IT relief – with full GDPR compliance.

Another example from the public sector shows how AI-supported knowledge management at ZAKB (waste management) answers technical questions in real time and enables application owners to develop new use cases much more quickly. The modular structure ensures that future enhancements can be implemented without major migration efforts.

In its analysis, ISG emphasizes that it is precisely this approach – modular architecture, interchangeable AI models, structured enablement – that significantly shortens the time-to-value from the initial proof-of-concept to stable production operation.

Looking ahead: What CIOs should tackle now

ISG identifies clear trends for the next three to five years: Event-driven automation will become the standard. Governance and cost control (FinOps) will increasingly merge into a common framework. And co-innovation – i.e., close collaboration between service providers and customers in mixed, results-oriented teams – will replace the traditional supplier-customer model.

For CIOs who want to take action today, ISG recommends a structured approach: Define your target vision, prioritize use cases based on a maturity analysis, and start with a pilotable minimum viable scope. This will allow you to achieve initial measurable results quickly without taking on the risk of a large, uncontrolled transformation.