In the area of ​​corporate AI, it is increasingly about the interaction of different interfaces.

Not a week goes by without even more intelligent language models, co-pilots or more powerful agents being announced to automate work in companies. The fact that artificial intelligence (AI) is becoming increasingly effective cannot be denied. But in practice, the majority of AI offerings sometimes offer companies little added value because they do not reflect the company's reality and ignore the fact that they cannot be easily controlled promptly.

An international manufacturing company trying to decide how to redirect its inventory in the event of supply chain disruptions needs more than just a simple, AI-generated answer. It must be able to find alternative suppliers, control inventory availability, view customer commitments and weigh financial trade-offs. And to forecast liquidity risks in volatile markets, CFOs need context that a simple chatbot cannot provide. All of these activities involve interconnected operational decisions involving dependencies, preferences, approvals, financial implications, and trade-offs that directly impact overall business operations.

In countless conversations that I have had with managers over the last year, the discussion has always shifted from AI capabilities to its feasibility in everyday work. While AI models are evolving rapidly, the real question is whether AI understands the business context in which it is used.

Too often in the discourse around AI, it is assumed that better models automatically lead to better business results. That is not the case. Companies are increasingly realizing that intelligent functions without an operational context – i.e. the processes, data, rules and guidelines that are used to control and protect companies – can initiate processes, but enable little progress. On the contrary, in some cases this can even lead to greater fragmentation and more risks.

An AI-generated recommendation can appear convincing, but it may ignore important dependencies in other parts of the system. And an AI agent can efficiently automate a workflow, but at the same time confuse the planning assumptions in another step. Companies are not lacking AI output, but rather AI systems capable of understanding operational impacts.

This is exactly where the real challenge for corporate AI currently lies. And to solve this problem, simply controlling processes is not enough. Context is needed.

Enterprise software has been the backbone of the global economy for decades. Financial systems, supply chains, procurement networks, workforce planning platforms, manufacturing operations and order fulfillment processes run through networked systems that capture not only information but also the logic behind company processes. These systems include process knowledge and data, governance structures, permissions, policies and economic relationships that have been collected over many years and are incorporated into every decision a company makes. They are the brains of every company.

In the age of AI, this business context is particularly valuable. Without this data, AI outputs remain well-intentioned guesses instead of well-informed judgments.

But when AI is embedded directly into operational processes, it can draw logical conclusions across all aspects of the company. This is changing the role of software in companies, as company systems are increasingly directly involved in the actual execution.

AI can identify risks earlier, coordinate responses across functions, provide real-time recommendations for action, and automate the execution of routine activities within defined frameworks. This does not happen in the form of isolated agents working separately from each other, but through intelligent functions that are linked to the economic and operational structure of the company itself.

Autonomy in a company does not mean that people are excluded from decision-making. Rather, autonomy means reducing the fragmentation and administrative burden that prevent companies from operating quickly, consistently and comprehensively. It is still people who set priorities, make important decisions and bear responsibility. However, AI can help coordinate and execute the operational processes related to these decisions.

As an example, let's consider the failure of a supplier that impacts a critical manufacturing component. Most modern AI systems can summarize the problem or predict the likely delay based on learned patterns. However, AI embedded in processes not only provides insights, but can coordinate and initiate actions. It can identify affected production schedules, analyze global inventory positions, evaluate alternative sourcing options, estimate financial exposure, highlight delivery risks to customers while suggesting actions for procurement, logistics, finance and customer operations.

AI not only automates work processes, but also offers completely new possibilities for the interaction between people and company systems. But not only that.

The more action-oriented AI becomes, the more crucial the systems that connect it to day-to-day operations and processes in practice become. Systems that can understand permissions, policies, dependencies, processes, financial impact and organizational responsibilities at the enterprise level are becoming more important than ever.

This development also affects how managers should deal with the topic of transformation.

So far, most companies have been experimenting with AI assistants, introducing pilot projects and automating isolated tasks. Only a few of them were able to actually increase their productivity, and even fewer were able to fundamentally realign their processes.

The companies that can emerge as pioneers in the next phase will take a different approach to artificial intelligence. You will connect intelligent functions directly to the operational systems where decisions have real business consequences. You will see that trustworthy, productive AI relies on context, data quality and process integrity, as well as extensive process knowledge.

Most importantly, these companies will understand that successfully leveraging AI is not just a technological shift, but a change management challenge. Real added value can only be achieved when AI agents, processes and people work hand in hand.

The future belongs to companies that find this balance: people set priorities and take responsibility, while intelligent systems coordinate processes and execute actions with precision. This enables companies to work more resiliently, productively and intelligently in an increasingly complex world.

Christian Klein is CEO of SAP SE.