Prism Data has raised USD 20 million in a Series A round led by Story Ventures and launched REFRACT, an AI transaction categorisation engine.
Cash flow underwriting company Prism Data has raised USD 20 million in a Series A round led by Story Ventures, with participation from FINTOP, Core VC, Obvious Ventures, 1st & Main Growth Partners, and Mana Ventures. Together with its USD 5 million seed round in 2023, the company has now raised more than USD 25 million. Prism plans to use the funding to invest in its technology, AI infrastructure, and data science, including a new generation of its CashScore models, and to expand across consumer and small business lending in the US and Canada, before eventually entering other markets.
Adoption of cash flow underwriting
Cash flow underwriting assesses creditworthiness using consumers’ bank transaction history, rather than relying solely on credit reports, which reveal little about how people earn, save, and spend. Prism’s technology is distributed through Equifax, Plaid, and LexisNexis, and the company works with fintech lenders and digital banks, as well as several large traditional banks that are implementing cash flow underwriting. According to Prism, the number of API requests lenders make for its solutions has increased more than 100-fold over the past two years.
Prism’s CashScore model was first used for consumer credit decisions by an FDIC-regulated institution in 2018. It has since been used across hundreds of lending portfolios covering tens of billions of dollars in loan originations, and has undergone model validation, compliance and regulatory audits, and independent fair lending testing. At Synchrony, cash flow underwriting helped enable nearly 50,000 additional approvals in 2025 for consumers who would otherwise have been declined, and Synchrony expects to open between 150,000 and 250,000 new accounts through cash flow underwriting in 2026, with these customers performing in line with its broader portfolio.
REFRACT categorisation engine
Alongside the funding, Prism has launched REFRACT, an AI-powered transaction categorisation engine trained on more than 30 billion bank transactions, which will underpin its next generation of scores and analytics. Rather than assigning categories based only on merchant names or keywords, REFRACT’s separate consumer and commercial models infer the financial purpose of each transaction, distinguishing earnings from borrowed funds, refunds from income, and purchases from transfers, even when the same payment platform is involved. According to Prism, REFRACT achieved double-digit percentage gains in categorisation accuracy over its previous models in benchmark testing.
REFRACT works with data from any Open Banking aggregator, financial institutions’ own deposit account data, and PDF statements across the US and Canada. It also monitors emerging transaction types and counterparties, including activity linked to new products such as prediction markets. Prism links the need for this approach to changes in how consumers earn and borrow, as gig work, self-employment, and products such as Buy Now, Pay Later, cash advances, and Earned Wage Access create income streams and obligations that traditional credit reports do not capture.
