New York-based Prism Data has raised $20 million in Series A funding to expand software that helps financial institutions evaluate creditworthiness using bank transaction data rather than relying exclusively on conventional credit reports.
The round was led by Story Ventures, with participation from FINTOP Capital, Core Innovation Capital, Obvious Ventures, 1st & Main Growth Partners and Mana Ventures.
The financing follows a $5 million seed round announced in 2023, bringing Prism's total funding to more than $25 million.
How Cash-Flow Underwriting Works
Traditional credit scoring relies heavily on information reported by lenders, including borrowing history, repayment performance and outstanding debt.
Those records can provide valuable signals about credit risk, but they do not offer a complete view of a person's current financial situation.
For example, a consumer may have limited credit history but receive a stable salary, maintain consistent savings and manage recurring expenses responsibly.
Another borrower may have a relatively strong historical credit score while experiencing a recent deterioration in income or cash flow.
Prism analyzes consumer-permissioned banking data to help lenders identify those differences.
Its platform examines income, spending, account balances, recurring payments and other transaction patterns.
That information can support a more current assessment of a borrower's ability to manage additional financial obligations.
CashScore and Refract
Prism's flagship product, CashScore, translates cash-flow information into credit risk predictions.
The model analyzes thousands of data points extracted from banking activity.
The company has developed additional models for different lending applications, including consumer and small business credit.
Alongside the financing, Prism introduced Refract, a system designed to improve transaction classification through AI.
The product attempts to distinguish money entering an account as genuine earnings from transfers, borrowed funds or refunds.
This matters because not every incoming bank transaction represents disposable income.
A transfer between a customer's own accounts should not automatically improve an income estimate.
Similarly, a loan deposit should not be classified as recurring earnings.
Refract is intended to improve the underlying data used for subsequent credit decisions.
Commercial Adoption
Prism reports working with three of the five largest fintech lenders in the United States and six of the ten largest digital banks.
Its products are also distributed through financial data and credit infrastructure companies including Equifax, Plaid and LexisNexis.
According to Prism, requests to its APIs have increased more than 100-fold over the past two years.
The company says CashScore models have been applied across lending portfolios representing tens of billions of dollars in loan originations.
Those figures indicate significant institutional adoption, although they do not disclose Prism's own revenue or profitability.
Founding and Competitive Position
Prism originated within Petal, a fintech company co-founded by Jason Rosen that developed credit products for customers with limited traditional credit history.
The technology was subsequently separated into an independent company.
Its competitors include established credit bureaus, financial data providers and specialized underwriting businesses.
The opportunity is not necessarily to replace traditional credit reports completely.
Instead, Prism wants lenders to incorporate additional information into existing credit decisions.
That strategy could be particularly relevant for consumers whose financial behavior is not adequately represented in traditional credit files.
The Risks of AI Underwriting
Credit models must meet demanding standards for accuracy, fairness, explainability and regulatory compliance.
Bank transaction histories can contain sensitive information, while patterns of income and spending may correlate with protected characteristics.
Lenders must evaluate whether new models improve predictive performance without creating unlawful discrimination.
Prism must also demonstrate that its models remain reliable during economic downturns, when consumer behavior changes.
A model that predicts risk accurately during stable conditions may perform differently during periods of unemployment or financial stress.
Prism's opportunity is to turn real-time financial information into better credit decisions. Its challenge is proving that cash-flow underwriting delivers more accurate and fair outcomes across different lenders, populations and economic conditions.