Spade, a data and AI platform for financial services, has launched Spade Predictions, a product suite designed to give banks and FinTechs access to transaction intelligence previously developed mainly by the largest financial institutions.

The suite is built around a foundation model trained on payment behaviour and matched with verified merchant data. Financial institutions can use the model as provided, fine-tune it for specific tasks or leverage Spade’s software, infrastructure and data embeddings to train their own models.

The launch comes as major banks and payment networks increasingly build proprietary AI models using their own transaction data. These projects require extensive customer histories, specialist research teams and significant investment, leaving smaller institutions with fewer resources to develop comparable capabilities. Spade aims to make transaction-based AI more accessible across the financial services industry.

Spade Predictions combines two components, one identifies the business behind a transaction using verified merchant records, while the other interprets spending behaviour to generate insights into customer needs and potential risks. This approach addresses a longstanding limitation of payment data, with card networks often restricting merchant names in transaction descriptions to 25 characters.

The technology can help financial institutions identify customers who may be suitable for particular products, anticipate future spending and detect potential fraud. Spade says it does not receive personally identifiable information, while models customised for individual institutions use those institutions’ own customer identifiers.

Early testing has produced promising results. Spade said a model fine-tuned for a US bank’s lending programme outperformed a conventional propensity model by 35%.

Source link