During a recent conversation on The Executive Outlook, Dr. Markus Schmidberger shared a warning that every enterprise leader investing in artificial intelligence needs to hear.
The biggest threat to enterprise AI is not technology. It is the weak foundation beneath it.
After leading data and AI teams across European companies for more than twenty years, he has seen organizations invest heavily in AI licenses, tokens, platforms and experimental tools. Employees write emails faster. Teams produce content more quickly. Processes become slightly more efficient.
Yet in many organizations, the business itself remains largely unchanged.
Customer problems are not solved differently. Products do not become more valuable. Revenue does not improve in a measurable way. Leaders spend more time discussing AI than examining whether it is creating meaningful outcomes.
He believes the current AI excitement may decline as companies begin questioning the return on their investments. This will not happen because AI has failed. It will happen because many organizations have approached it without a clear business purpose.
Too many leaders are building an AI adoption strategy for enterprises before understanding the problems they want to solve, the data available to support those problems and the organizational capabilities required to turn technology into value.
For him, successful AI adoption is not simply a technology decision. It is a leadership challenge involving data, culture, trust and business alignment.
He developed an interest in data long before artificial intelligence became a boardroom priority.
As a student, he created Excel tables to track his school marks. He wanted to understand how his grades were developing and predict the results he might receive at the end of the year.
That early instinct became more serious during his doctoral work in biocomputer science, where he analyzed cancer data at a global pharmaceutical organization.
Initially, the work involved simulations using data from ten people. At that scale, researchers could still review individual records manually. They could examine the information closely, identify unusual results and apply human judgment to each case.
Then the analysis moved from a standard computer to a supercomputer. The research expanded from simulations involving ten people to simulations involving ten thousand.
Manual review is no longer possible.
The team had to rely on algorithms, quality checks, testing processes and a clear data strategy. Excel was no longer sufficient. Human observation could not support the volume or complexity of the information.
That transition shaped how he now thinks about enterprise AI.
A small experiment may work through individual effort and manual supervision. Scaled execution requires structure, governance, reliable data and confidence in the systems making decisions.
This is where many enterprise AI programmed begin to struggle. Organizations successfully run an AI pilot, but they lack the foundation required to make it part of everyday business operations.
Every successful AI adoption strategy for enterprises begins with a simple question.
What data do you have?
When Dr. Markus enters an organization, the first answer is often immediate.
The data is in the data warehouse.
But that answer is not enough.
What specific data is stored there? Which customers, products, transactions, or business entities do they represent? How much historical information is available? Which periods are reliable? Where are the quality gaps? Who owns the information? Can employees' access and understand it?
At this point, many organizations have become uncertain.
He believes that a company able to answer these questions confidently is already ahead of most businesses. The challenge is not that companies lack data. The challenge is that they often do not know exactly what they have, what it means, or whether it can support a reliable decision.
This matters because data describes how a business operates.
It reveals how potential customers become paying customers. It shows how people use a product, where marketing performs well, where customers leave, how money flows and where operational problems appear.
When AI has access to reliable business data and the context surrounding it, it can help improve processes, products, decisions, and customer outcomes.
Without that context, AI is being asked to make decisions about a business it does not fully understand.
Enterprise data management provides the definitions, ownership, history, and quality standards needed to make data usable. A data catalogue may help teams understand what information exists. Clear semantic definitions can help both employees and AI systems interpret the information correctly.
The principle is straightforward.
Before asking what, AI can do, leaders must know what data their organization has and whether that data can be trusted.
A strong data foundation is necessary, but it is not enough.
An organization may build an advanced data warehouse, create a modern data lake and hire skilled specialists. But if the rest of the business cannot access or understand the information, the investment will struggle to create a wider value.
This is why data literacy is central to AI adoption.
He recalls working with a large technology marketplace that managed a significant amount of data. Within the organization was a small team of three people responsible for arranging travel for senior management.
The travel team was not considered a major candidate for transformation. Its work appeared administrative, and much of its information was managed through Excel.
The team was eventually given access to the company’s data lake through a business intelligence tool. Once the employees could examine their information more clearly, they discovered that the organization was losing a considerable amount of money through incorrectly booked hotels and inefficient travel processes.
They changed the booking approach, introduced a better system and reduced annual travel costs by approximately thirty percent.
The opportunity was not discovered by senior leadership or a specialized AI team. It was discovered by employees who understood the daily process and were finally given the tools to analyze it.
That example changed how he viewed data transformation.
The most valuable opportunities are not always visible from the boardroom. They are often found by employees who understand the delays, exceptions, repetitive tasks, and operational problems that leaders rarely see directly.
Data literacy allows those employees to participate in transformation instead of waiting for a central technology team to solve every problem.
For enterprise leaders, data literacy should not be treated as a one time training activity. It should be viewed as part of the organization’s competitive strategy.
The companies that create value from AI will not be those with only the strongest technical teams. They will be those where finance, operations, marketing, product, customer service, and support teams can identify problems and discuss them through a shared understanding of data.
One of the largest gaps in enterprise AI comes from confusing assistance with transformation.
Many organizations use generative AI to draft emails, write social media posts, summarise documents, or support research. These activities can save time and improve personal productivity.
But they rarely change the underlying business process.
In this form of AI use, an employee enters a prompt, receives an answer and remains responsible for every meaningful action that follows.
The next stage is automation. AI helps classify information, generate responses, route requests, or reduce the amount of manual work involved in a process.
The more significant shift begins with AI agents.
A typical customer support chatbot retrieves answers from a frequently asked questions page. When the problem becomes complicated, it transfers the customer to a human support employee. It provides information, but it does not solve the issue.
A genuine AI agent can take action.
It may update information in a system, change a record, complete a task, or move a request through an entire process. Rather than only explaining what should happen, it helps make it happen.
This distinction matters for large enterprises as well as organizations exploring AI agents for small businesses. The size of the organization may change the complexity of the system, but the principle remains the same.
Business value increases when AI moves beyond producing answers and begins improving outcomes.
However, greater autonomy also introduces greater responsibility. An AI system that can change information or make decisions requires reliable data, clear permissions, monitoring, compliance controls and defined points of human intervention.
This is especially important in healthcare, financial services and other regulated industries. He advises organizations not to begin with their most complex or sensitive use cases. The opportunity may be significant, but the risks and control requirements are also higher.
Leaders should begin with simpler processes where the organization can learn, build confidence and measure value before expanding into more critical areas.
Many organizations say they want AI to take on more work. At the same time, they attempt to review and control every action it performs.
He describes this behavior as micromanaging AI.
The concern is understandable. AI may produce incorrect information, misunderstand context, or generate a confident response based on incomplete data. Leaders cannot simply hand over critical responsibilities without boundaries or accountability.
But refusing to delegate anything also limits the value AI can create.
He explains this through the relationship between an executive and a trusted assistant.
An executive can review every calendar of invitation, travel booking, meeting notes and administrative decision made by the assistant. But that level of supervision may take more time than completing the work directly.
The assistant becomes valuable when trust is established. Routine responsibilities are handled independently, while only the most important decisions are brought to the executive.
Organizations must develop a similar relationship with AI.
The goal is not blind to trust. It is informed delegation.
To trust AI responsibly, leaders need to understand what the system can do, what information it can access, which actions it may take, how performance will be monitored and when a human must intervene.
Trust is created through clear controls, reliable data, transparency and experience.
Organizations that never move beyond constant supervision may continue paying for AI without receiving its full value. Those that build responsible trust can begin redesigning how work is completed.
He offers a direct recommendation to leaders in developing an AI adoption strategy for enterprises.
Do not create a business strategy, a data strategy, and an AI strategy as three disconnected documents.
Create one business strategy that explains where the organization is going, which customer or operational problems it wants to solve and how AI can support those goals.
Data should be recognized as the foundation required to execute that strategy.
A separate implementation plan may still be necessary. It can define data priorities, AI use cases, responsibilities, milestones, controls and measures of business value. But every initiative must connect to one shared direction.
When strategies are separated, teams often optimize for different outcomes.
The business focuses on growth. The data team focuses on infrastructure. The AI team focuses on models and experimentation. Each group may perform well within its own area while the organization as a whole fails to create value.
Alignment must begin with leadership.
Before executives approve a large AI programmed, Dr. Markus recommends that they gain practical experience with technology.
In one session, an executive team used a few prompts to build a version of the classic Snake game. The goal was not to turn the chief executive, chief financial officer, or chief data officer into programmers.
The goal was to help them experience how quickly the nature of building and problem solving has changed.
Leaders cannot guide an AI transformation through presentations alone. They need enough practical understanding to ask better questions, recognize realistic opportunities and challenge investments driven by hype instead of value.
Even the strongest strategy will fail if employees do not use what the organization builds.
Many employees hear predictions that AI will replace jobs, reduce teams, and eliminate roles. When leaders introduce a new AI tool, people may view it as a threat rather than support.
Fear creates resistance. Employees avoid the system, continue working through familiar processes, or hesitate to share ideas.
Leadership communication must therefore be clear.
Employees need to understand that AI is being introduced to reduce repetitive work, support better decisions and create more time for problems requiring human knowledge, relationships, creativity and judgment.
When organizations communicate this purpose honestly, employees become more willing to experiment. They start identifying use cases from their own work.
This is where AI literacy becomes essential.
For years, organizations have discussed the gap between data teams and business teams. He believes a similar AI to the business gap could emerge.
Technology teams may understand what AI can do, while business teams struggle to connect those capabilities with customer, operational, or financial outcomes. At the same time, AI systems may lack the business context required to produce useful answers.
AI literacy gives employees the confidence to discuss the technology, understand its limitations, recognize realistic applications, and contribute ideas without fear.
Near the end of the conversation, He suggested a simple experiment.
Ask an AI tool to generate a random number between one and one hundred. In his experience, the response frequently lands on seventy-three or forty-two.
His point was not about the numbers themselves. It was about the statistical nature of AI.
AI generates outputs based on patterns found in its training data. This allows it to process large amounts of information and identify connections at a scale no individual can match.
But it can also repeat common patterns, miss important context, or provide a confident answer that is not appropriate for the situation.
Leaders must avoid two extremes.
They should not reject AI because it is imperfect. They should also not accept every output because it appears intelligent.
Trust AI to support the work but never switch off human judgment.
The central lesson from Dr. Markus is clear.
The AI adoption strategy for enterprises does not fail because technology lacks potential. It fails when organizations invest before understanding their data, aligning leadership, educating employees and defining the business value they want to create.
The organizations that lead will know what data they have. They will make that data accessible. They will build data literacy and AI literacy across the business. They will move beyond basic chat tools toward systems that improve outcomes. They will create responsible ways to trust AI.
Most importantly, every AI investment will connect to a clear business strategy.
The AI business gap is beginning to form.
Enterprises that close it now will not simply use AI more often. They will use it with stronger judgment, clearer purpose and a more direct path to measurable value.
Want to hear more conversations with leaders building the foundations for responsible AI and data transformation? Explore more insights on The Executive Outlook.
Listen in for business strategies and interviews with business experts to help you build a business and a life that you love.
About Us
