Samenvatting
Becoming an AI-first learning organization is not about adding AI to existing processes, says UCB’s Aitor Larrabe. It requires rewiring how learning and talent development work, with the ambition of making all HR services available to all employees, all the time. Technology can personalize support and extend its reach, while people remain essential in the moments that matter. Trust, responsible data use, experimentation and the ability to scale are crucial. For learning professionals, this is a defining moment: their role is shifting from supplying standard programmes to orchestrating personalized growth experiences. Larrabe’s advice is simple: start experimenting, learn from experience and focus on helping people grow in every possible direction.
“What if we could offer all HR services to all employees, all the time?”
Adding AI to an existing learning process does not make an organization AI-first. For Aitor Larrabe, Head of Talent Acquisition and Learning at UCB, the real opportunity lies in fundamentally changing how organizations help people grow. That means offering more personalized support at a much larger scale, while preserving human interaction in the moments that truly matter. It also requires trust, experimentation and learning professionals who understand data and technology as well as people.
Learning across worlds
Aitor Larrabe combines his role at UCB with teaching at IE Business School and activities supporting people with rare diseases. He co-founded Foundation 29, an organization that develops technology to improve the diagnosis of rare diseases. Moving between the corporate, academic and non-profit worlds gives him different perspectives on learning, innovation and human development.
“The best way to learn something is to teach it,” he says. “What I learn in one environment, I can use in another. Teaching executives from other companies also gives me insights into what is happening in the outside world.” Over the course of his career, Larrabe has worked in sectors ranging from technology and construction to pharmaceuticals. “I have seen different flavours of companies,” he says. That experience has taught him that there is no universal model for learning and development. What works depends on an organization’s culture, leadership, business needs and readiness for change.
More than adding AI
Becoming an AI-first learning organization does not mean placing a chatbot on top of an existing platform or using generative AI to produce more content. For Larrabe, AI matters because it creates opportunities to offer services that were previously impossible, or at least impossible at scale. “It is less about AI and more about changing the way we work by using AI,” he says. “Technology gives us the opportunity to offer services that we were not able to offer before.”
He captures that ambition in one question: “What if we could offer all HR services to all employees, all the time?” Today, most organizations provide some HR services to some employees at certain moments. Personal attention remains scarce and is often reserved for particular populations or pivotal career moments. An employee who wants career advice may need to schedule a conversation with an HR or talent partner, wait several weeks and then explore a complex question within only one hour. Even in organizations that invest heavily in people development, that model cannot offer everyone the depth and frequency of support they may need.
Deeper reflection
An AI agent could help employees reflect on their interests, strengths, working preferences and possible career paths. It could ask questions, challenge assumptions and allow someone to explore different scenarios over several conversations. Employees could then enter a human conversation with greater clarity. “A career question often requires profound reflection,” Larrabe explains. “You need to provide a lot of context, be very specific and be incredibly honest. That is difficult to do in a single conversation of one hour.”
Suppose someone working in HR is curious about procurement. An agent could help that person explore why the field appeals to them, how it fits their preferences and which skills they would need. Technology might subsequently identify an internal project through which they can test that interest. At another point, a human adviser may be needed to discuss the actual opportunities within the organization. The technology does not need to take over the entire journey. Its value may lie in improving the quality of reflection and allowing human expertise to be used where it adds the most value.
Moments that matter
Larrabe therefore rejects the binary choice between people and technology. The same employee may prefer a digital tool in one situation and a person in another. Some people may even feel more comfortable discussing sensitive questions with a system, while others will refuse to talk to what they perceive as a robot.
He compares this evolution with navigation systems and online banking. Not everyone wanted a GPS when the technology first appeared. Some drivers preferred to trust their own instincts. Today, most readily accept real-time directions from a system while still enjoying the company of the person sitting next to them. Banking followed a similar path: visiting a building and speaking to an employee once felt indispensable, while many people now handle most transactions digitally.
Preferences can shift quickly when technology becomes familiar and genuinely useful. But Larrabe does not assume that every HR interaction will follow the same trajectory. “At UCB, we want to preserve human interaction in the moments that matter,” he says. “The difficulty is that those moments are not the same for every person or in every situation.”
The challenge is to offer meaningful alternatives and learn which combination produces the best outcome: a human interaction, a technology-enabled service or a human professional whose judgement is enriched by better data.
Trust comes first
Before choosing technologies, UCB articulated why it wanted to transform learning. Development needed to become more personal, more closely embedded in the flow of work and more relevant to the business. That ambition also required clear principles. The most important is trust. “Trust is the bridge between the known and the unknown. If you destroy that bridge, you may never get a second opportunity.”
If employees use an avatar or AI agent to reflect openly on their strengths and development needs, they need to know how that information will be used. An organization cannot present a tool as a safe environment for development and later use its output to decide who should leave. That would immediately undermine the credibility of the system and of future initiatives. Privacy and data use are therefore not issues to address after implementation. They need to be part of the design from the beginning. Employees will also make different judgements about the exchange. Some are comfortable sharing data if they receive a useful and personalized service in return; others are far more reluctant. Both positions need to be taken seriously.
Experiment and scale
In a highly regulated biopharmaceutical environment, experimentation needs clear boundaries. UCB has developed an approach that allows employees to test ideas without every team reinventing the same solution. People with a promising use case can receive support to experiment. When an idea works, the aim is to turn it into a reusable solution that can be scaled across the organization. UCB calls this model its AI-first factory, a name that Larrabe admits occasionally provokes discussion. “We see it as R&D within Learning and Development. We try something and assess whether it looks promising. Then we scale it, and if it scales successfully, it becomes a product that others can use.”
The principle is straightforward: if a solution already exists for a particular use case, employees can access it through the factory. If it does not exist, there is room to conduct R&D, pilot an idea and learn from the result. Talent Acquisition was one of the first domains UCB selected for transformation because the technology was sufficiently mature and the necessary data was available. Other applications are still being explored through smaller experiments. The model combines bottom-up creativity with the discipline needed to turn isolated pilots into organization-wide solutions.
Rewire the work
Most large organizations will eventually have access to similar AI technologies. Larrabe believes that the real difference will lie in the quality of their data, their understanding of the context and their ability to redesign processes. “AI-first is not the status quo plus AI. It is about changing the way you operate to achieve the best possible outcome.”
That may mean improving a process with AI, but it can also mean abandoning a process that no longer makes sense. Automating a weak learning process does not make it a strong one. Continuing to invest in an outdated approach also diverts money and energy from solutions that may create more value.
This is why UCB increasingly talks about innovation rather than AI alone. Technology is only one component. Organizations also need the right data flows, redesigned processes and the trust required to use the resulting services at scale.
Scale needs to be considered from the outset. A pilot involving two people can become a test with one hundred, then one thousand and eventually ten thousand employees. If that possibility is absent from the initial thinking, the organization may never create the data and processes required to move beyond the pilot stage.
Signal and noise
Implementation is difficult because AI is not the only transformation taking place. Employees are simultaneously confronted with new systems, processes and ways of working. Travel services, workplace technology and many other parts of the employee experience may all be changing at once. “At UCB, we talk about separating the signal from the noise,” Larrabe says. “There is so much happening internally and externally that it becomes difficult to identify what is genuinely important.”
Expertise is another challenge. Experienced specialists are in high demand, while many vendors are still developing their own capabilities. Organizations often work with enthusiastic people who have never implemented this technology before, partly because the field is so new that few people have.
Adoption is also shaped by individual preferences and broader social concerns. Some employees worry about privacy or want to spend less time in front of a screen. Others embrace the convenience of digital support. An organization has to navigate all those different ‘flavours’ while the internal market itself may change rapidly.
A defining moment
Larrabe sees this as a career-defining period for learning and HR professionals because their work influences moments that can profoundly affect people. An unsuccessful application for a dream job is not simply another step in an HR process. The way an organization handles that decision can damage someone’s motivation or help them move forward with new energy.
“We sometimes look at these experiences from the supply side: the message has been sent, so we tick the box,” he says. “But for that individual, it may be one of the most important moments of the past ten years.” To understand the weight of such moments, Larrabe offers a deliberately provocative thought: “An HR person should probably be fired at least once in their life, perhaps even unfairly.” His point is that experiencing the bitterness of such a decision changes how someone approaches it as an HR professional. “If it has never happened to you, it is difficult to know what it feels like. In these moments, we may kill or boost someone’s motivation without even noticing it.”
Technology can provide faster and better-informed support, but it also increases the responsibility of those designing the experience. HR and learning professionals can influence someone’s confidence, career and professional identity. As Larrabe puts it, borrowing from Spider-Man: “With great power comes great responsibility.”
New L&D skills
More personalized learning requires a much deeper understanding of the audience. A broad category such as middle managers is no longer sufficient when an organization wants to serve thousands of managers with different needs, contexts and preferences. Learning professionals will need to understand segmentation, data flows, new formats and the expanding possibilities of technology. They also need to assess whether an intervention improves performance rather than simply count participation or content consumption. “Being strategic is not about sitting back, smoking a cigar and thinking about things,” Larrabe says. “It is about creating the data points that allow you to make better decisions and have more impact.”
The role shifts from supplying standard programmes to understanding demand, creating meaningful options and orchestrating the right combination of human and technology-enabled support. Even experienced specialists will need to keep learning as simulations, avatars and AI agents create entirely new ways to practise and develop skills.
Growth hacking people
Larrabe’s role brings together Talent Acquisition, Learning and Talent Management. He regards them as different parts of one mission: helping people grow. “My department has a traditional name, but in reality, it is the ‘helping people grow’ department. You could even call it the growth hacking department.”
Growth does not automatically mean promotion. It may involve moving into another role, developing new skills or becoming better at the job someone already performs. “Someone can stay in the same role for twenty years and still grow. If you go to work every day wanting to do better, you will become better and better at what you do.”
Employees do not experience recruitment, learning, internal mobility and talent management as separate HR processes. They experience one career. Technology creates an opportunity to connect the support surrounding that career and make it more personal, accessible and continuous.
Start trying
Larrabe does not believe in one blueprint for the AI-first learning function. The right approach depends on the organization, its culture, its leaders and its maturity. Some companies need more room for innovation. Others first need greater operational discipline. Even the most ambitious learning offer will struggle if managers see development as a distraction from everyday work. Creating appetite for growth without being able to provide meaningful opportunities can be equally damaging.
His most practical advice is simple: start experimenting. “You will not improve your offering, your processes or your technology if you do not try things. Test them and experience them yourself, even with use cases that may not seem particularly important. Going through that experience gives you a completely different understanding.”
For Larrabe, becoming AI-first means rethinking how the organization helps people grow and using technology to make that support available to more people, more often.
===
Meet Aitor at AI for Learning Day
Aitor Larrabe will speak at AI for Learning Day on December 3rd 2026 in Mechelen. In his keynote Turning into an AI-first learning organization, he will share how UCB is rewiring learning, where AI creates genuine value and what this transformation means for learning professionals.
