Commerzbank launches an AI Academy to build AI skills across the bank

You have probably sat through a vendor pitch that promised to "democratise AI" or "upskill the workforce at scale." The slides were polished. The roadmap looked plausible. Then you returned to your desk and found that nobody could write a useful prompt, half the team feared the compliance risk and the other half had already built a shadow ChatGPT habit around your data-loss-prevention rules. Commerzbank decided that the gap between ambition and capability was a design problem, not a communication problem, and built the AI Academy to close it.

The academy is a structured learning environment that teaches employees and early-career hires how to use AI tools safely, effectively and responsibly in daily work. It does not train data scientists. It trains bankers, analysts, advisors and project managers to integrate AI assistants into the processes they already own. The curriculum is organised into learning paths that match role and experience, starting with a User Basic Track that covers fundamentals: what AI can do, how to write prompts that produce useful results and how to move from experimentation to consistent application. Participants learn to use AI assistants strategically, develop custom agents and embed AI into workflows so that it becomes a natural part of the way they work, not a side experiment.

The programme also includes a short-form series called AI Espresso, fifteen-minute sessions that deliver current trends, best practices and tool updates. The format is designed to fit into a working day without requiring calendar Tetris or approval chains. Yasmin Lutz, Head of Learning & Transformation at Commerzbank, says the academy is "the starting point for a vibrant community where we learn from one another, share knowledge and explore new paths together." The community framing matters because it signals that the bank expects learning to be continuous and peer-driven, not a one-time certification event.

What responsibility means in practice

The academy does not treat responsibility as a compliance checkbox. It teaches employees why data quality matters, how sensitive information must be protected, where the limitations of AI lie and which regulatory and ethical requirements apply. The curriculum is explicit: even in an increasingly digital world, people make decisions, support customers and drive innovation. AI is a tool that amplifies judgment; it does not replace it.

That framing is deliberate. Financial services operate under stricter data-protection and conduct rules than most industries, and the cost of a poorly designed prompt or a mishandled customer interaction is not just reputational. It is regulatory. The academy addresses that risk by embedding responsibility into the learning design, not as a warning slide at the end but as a principle that shapes how employees are taught to interact with the technology from the first session.

The tools employees learn to use include Microsoft Copilot, ChatGPT and Google Gemini. The academy does not prescribe a single platform. It teaches people how to choose the right assistant for the task, how to evaluate output critically and how to integrate AI into processes that already have quality controls, approval workflows and audit trails. The goal is not to automate judgment but to give employees more time to apply it by removing low-value work.

Why this matters for L&D

Most enterprise AI training falls into one of two traps. The first is the vendor roadshow: high on vision, low on application, with no follow-through after the session ends. The second is the compliance module: a mandatory click-through that teaches people what not to do but gives them no practical skill to apply on Monday. Commerzbank's model avoids both by making learning continuous, role-specific and community-supported.

The academy also solves a structural problem that many learning teams face: how to scale capability-building without creating a bottleneck in the L&D function. By organising content into self-serve tracks, offering short-form updates and building a peer-learning community, the bank distributes the teaching load across the organisation. Employees learn from one another, share what works and surface new use cases that the curriculum designers did not anticipate. That feedback loop makes the programme adaptive, not static.

The other strategic choice is the focus on young talent. Early-career employees are often the fastest adopters of new tools and the most willing to experiment, but they also have the least institutional knowledge about risk, compliance and customer impact. By giving them structured training that combines capability with responsibility, the bank builds a cohort of AI-literate professionals who understand both the technology and the context in which it must be used. That cohort becomes the internal capacity for the next wave of transformation, without needing to hire it from outside.

If your organisation is still treating AI training as a one-off event or a vendor-led workshop, you are building capability that will not compound. Commerzbank's approach shows what happens when you design learning as infrastructure: role-specific paths, continuous formats, peer networks and responsibility embedded from the start. The result is not a certificate. It is a workforce that can use the tools, evaluate the output and make the judgment calls that matter when the technology cannot.

Sources:

• Commerzbank, 15 October 2026