Banco Santander continues to build on its artificial intelligence (AI)-powered automation model by moving beyond its use for specific tasks towards more complex operational capabilities that can be rolled out across the globe. Several technologies are supporting this drive, including that created by Spain-based startup Maisa.
This initiative is part of One Transformation, Santander’s model to simplify and standardize processes, share technology and capabilities between markets, and harness its global scale. AI is one of the major drivers of this transformation, with the aim to generate over 1 billion euros in business value by 2028.
Santander is developing AI systems that can run processes by combining reasoning with execution, with the help of Maisa. This technology keeps a record of each run (the information used, steps followed, and how the outcome was obtained), which means processes can be traced, audited and reproduced in keeping with regulatory requirements.
This traceability also enables the bank to train its own specialized models and roll them out across its infrastructure under full governance and maximum cost efficiency. In turn, the platform allows for the reuse of these capabilities in other processes and markets. The goal is to build once and scale what works. Santander already has hundreds of use cases in production with Maisa support across its main markets in Europe and the Americas.
This model also draws on the expertise of Santander’s own teams, which can develop automation within a common technology and governance framework. This brings automation capabilities closer to the people who understand each process best.
The aim is not to automate for the sake of automating, but rather to spot where AI can simplify processes, make them run better, and free up time for added value tasks, as well as scaling lessons learned to generate greater impact.
Maisa is a Spain-based company founded in 2024. Its technology focuses on the traceability of decisions (the basis for training specialized process models) and on preventing errors in AI known as “hallucinations” (i.e. when an AI-powered system provides incorrect or made-up information).
