Banks are facing not an AI problem, but a data platform problem.

The CBA Live 2026 conference discussed the challenges of implementing artificial intelligence (AI) in banks. The main barriers to scaling AI solutions turned out to be insufficient data infrastructure and a weak data management system. Despite having extensive customer data, the lack of integration between systems makes it difficult to use them effectively.

The innovations presented at CBA Live 2026, including Databricks Lakehouse and Unity Catalog, are aimed at improving data quality and monitoring models. Participants noted that successful banks are not those with the most advanced AI technologies, but those that have created a clean and real-time accessible database.

Experts found that many banks face challenges in risk management and regulatory compliance. For example, a credit scoring model trained on one dataset may perform poorly when the composition of clients changes. This is because the level of data quality required for AI significantly exceeds the expectations of many teams responsible for compliance.

Areas such as debt management and personalized financial offerings demonstrate how the right data infrastructure can enhance customer engagement. The application of AI to predict the timing of bank application submissions with 85% accuracy illustrates how early signals can change resource allocation in collections.

Successful financial institutions focus on creating data, to which AI functionalities are then added. By updating their internal data systems, they were able to develop AI models, which in turn improves customer service. This evolution is becoming necessary for competitiveness and enhancing service levels.

The importance of continuous data monitoring to prevent AI model degradation was also emphasized at CBA Live 2026. Successful AI application requires a deep understanding of customer data and integration with business processes, highlighting the significance of a quality database for the future of banking services.