When ChatGPT launched a global corporate scramble in late 2022, Walid Mehanna was already deep into rebuilding the digital backbone at the world’s oldest pharmaceutical company. As a data set officer in Germany Merck KGaA (also known as Merck Group in the US and unrelated to US pharmaceutical firm Merck) at the time, Mehanna had spent two years overhauling the company’s data and analytics systems for its 62,000 employees. About eight months into the GPT era, his title was promoted to chief data officer and AI officer. The creation of that role, he said, “was an evolution by design, not a reaction to a development, buzz or trend.”
The roots of Merck KGaA stretch back to the 17th century, when pharmacist Friedrich Jacob Merck laid the foundations for the family business in Darmstadt, Germany. Today, its business includes pharmaceuticals, medical device manufacturing and electronics. Leading an AI strategy within a 350-year-old company is a unique kind of challenge. Mehanna, who was born in Egypt, compares it to building a pyramid.
At the core are everyday AI tools that improve personal productivity across the workforce. The goal is to build digital fluency, save time and reinvest these benefits into growth. The middle tier—where the company is today—is focused on incorporating AI into core workflows, such as research and development, supply chains, and commercial operations over the next several years. At the top of the “pyramid” is product AI, where machine learning becomes part of what the company sells, helping accelerate drug discovery and innovation.
“Philosophically, AI at scale is not a technological challenge,” Mehanna said. “You still have to do your homework on the technology side, but real transformation is about leadership. It happens when strategy, culture and ambition come together with clear commitment.”
Choosing a local start-up partner
Merck KGaA was an early mover into generative AI by opening an internal platform called MyGPT to its employees in June 2023. Originally built in-house, the tool gave staff a safe space to experiment. But Mehanna soon saw the limits of relying on a single structure.
Within a year, Merck KGaA replaced its original configuration by merging with LangDocka then Berlin-based startup. At the time of their initial conversations, LangDock had four customers and approximately $50,000 in annual recurring revenue.
Choosing a new European company can seem political in today’s AI landscape. Mehanna says it wasn’t.
“The thinking behind the collaboration was about optionality, speed and sovereignty, not the sentimentality of working with a German startup,” he said. “LangDock gave us the opportunity to build something fully GDPR compliant that we could host in our own environment. We were able to work with the company to achieve enterprise-level security while maintaining the agility of a startup.”
The setup adds a buffer between employees and key AI providers, giving Merck flexibility as models evolve. “It was important to us to create a flexible, model-agnostic user layer that gave us access to the best models in the world, regardless of who made them, without creating vendor lock-in,” Mehanna said.
Building AI protective arms like Mercedes brakes
Deploying AI in a global workforce the size of Merck KGaA requires navigating strict regulatory and labor relations frameworks, particularly in Europe. Mehanna chose to build governance into the strategy from the start, working closely with the company’s works council.
Since deploying AI, Merck KGaA has internally generated over 12 million queries, a goldmine of data containing insights into how employees work, what’s on their minds and what they need AI to help them with. Mehanna said the company analyzes these requests strictly on an aggregate level and never monitors individual employee activity.
“We learn from patterns in requests, never from individual employees or their usage,” he said. “All use is privacy protected and we do not track individuals.”
Prior to joining Merck KGaA, Mehanna served as Chief Data Officer at Mercedes-Benz for five years. He often compares AI ethics to the brakes on a high-performance car: easy to overlook, but essential.
“My point of view is that I want to have the best brakes in the world so I can go fast. A Mercedes-AMG, Porsche OR Ferrari it has some of the best brakes because it also travels at very high speeds,” Mehanna said. “Governance and ethics work the same way. I want better governance and ethics to be in place because the right handrails allow us to move quickly while staying safe and secure.”
Putting this philosophy into practice, Merck KGaA created a Digital Ethics Advisory Panel guided by five principles: autonomy, benevolence, non-maleficence, fairness and transparency. Rather than simply approving or rejecting projects, the panel pushes teams to identify risks early and design safeguards before launch.
“Our outside experts ask tough questions and make us think carefully about the consequences of possible actions. We then determine what handrails are needed to proceed in a safe manner and in line with our values,” explained Mehanna.
Why models are no longer on the cutting edge
Like many technology leaders, Mehanna’s view of the market has matured alongside technology. He no longer believes that the entrepreneurial race will be decided at the level of the fundamental model, as the frontier models constantly bounce off each other.
“Two years ago, I was pretty convinced that the race for general artificial intelligence or enterprise AI would be decided at the model layer—that whoever had the best model would win. I’ve definitely changed my mind about that,” Mehanna admitted.
“The models are still decisive, but they cross each other every few months. So they are not a sustainable advantage,” he added. “I believe the most lasting advantage lies in your context, data, processes, people fluidity and, most importantly, the trust you build with your organization and your customers.”
This change has changed the way Merck KGaA builds its systems. A top-down semantic data layer across the company proved too slow. Instead, teams now integrate data incrementally – process by process and use case by case.
On the inevitable question of AI’s impact on work, Mehanna dismisses the notion that AI will simply eliminate human roles. Instead, he rewrote the job requirements.
“We don’t see roles disappearing. We see roles evolving,” Mehanna said. “Artificial intelligence becomes a dynamic tool for accelerating and extending work. The leader of the future will also need to lead a hybrid workforce composed of humans and agents.”
Reflecting on leading through industrial change, Mehanna references an unexpected source: Arnold Schwarzeneggermanagement guide, Be Useful: Seven Tools for Life. Through diverse careers in athletics, entertainment and politics, the fundamental constant remains clear.
“The basic underlying principle is to always be helpful, and that resonated deeply with me,” Mehanna reflected.
