Generative AI has proved powerful at language but weak at prediction, prompting banks and technology firms to pursue new approaches. Companies increasingly rely on structured data to forecast churn, demand, and credit risk—tasks where large language models fall short.
TD Bank has developed a tabular foundation model called Prism through its AI unit, Layer 6, to address the challenge. Trained on a year of transaction, account, and credit bureau data, Prism allows staff to forecast customer behaviour without building separate models for each task. Executives say the approach scales predictive analytics more efficiently across the organization.
Want to know more? Check out the source code The Logic.











