Propel adds bank-transaction AI to credit decisions, with Plaid supplying data
The lender estimates credit access could increase by up to 15%. That is a company forecast, not a measured gain, and U.S. expansion is expected in early 2027.
Loading page…
The lender estimates credit access could increase by up to 15%. That is a company forecast, not a measured gain, and U.S. expansion is expected in early 2027.
Listen to this story
Propel’s proprietary cash-flow underwriting system has been operating since late Q3 2026, using six months of account history and Plaid’s permissioned financial data to assess applicants beyond credit-bureau files. Propel estimates the approach could expand credit access by up to 15% while keeping risk thresholds unchanged, but has not disclosed approval comparisons, sample size, or repayment results. A U.S. rollout is planned for early 2027, making that expansion the next test of the system’s reach.
Propel says its proprietary AI underwriting dates to 2015; it and bank partners have since underwritten more than two million loans and lines of credit.
Co-founder Jonathan Goler says the model uses time-series deep learning to identify repayment patterns in applicants’ cash flow.
Propel describes personalized lending and a broader product range as possible future uses, not offerings launched with this announcement.
Consumers overlooked by conventional credit checks could gain another route to approval through their bank transactions. Propel Holdings on November 18, 2026, announced an AI model that assesses six months of account history, supported by a Plaid partnership. Propel says early results show more consumers can qualify without changing its risk thresholds.
The announcement describes a system already in use, rather than a launch still to come. The enhanced cash-flow model began operating in late Q3 2026. Propel estimates it could increase access to credit by up to 15%, but that figure is a forecast—not a disclosed increase already achieved.
Underwriting is the process of assessing whether someone qualifies for credit. Propel built the model entirely in-house and says it analyzes six months of bank transaction history in seconds. Its stated purpose is to give Propel and its bank partner a fuller view of an applicant’s finances than traditional credit bureau data alone.
Plaid supplies consumer-permissioned financial data for that assessment. Its contribution includes insights into cash flow, income patterns and financial obligations. Those inputs support Propel’s own AI models; the announcement describes Plaid as the financial-data partner, while the underwriting model remains Propel’s proprietary technology.
Jonathan Goler, Propel’s co-founder and chief risk officer, says the system combines open banking data with time-series deep learning models—AI designed to examine patterns over time. He frames the benefit as understanding a consumer’s current financial circumstances and finding patterns predictive of repayment, rather than relying on a credit score to tell the whole story.
Plaid is helping Propel close an information gap with access to real-time cash flow insights and a network-level view of a borrower's financial obligations, so the team can manage risk and decision borrowers with confidence.
Michelle Young, Credit Product Lead at Plaid
Propel’s performance claim has two parts: approving more consumers and doing so at the same risk thresholds. The company presents the model as a way to identify creditworthy applicants conventional methods may miss, not as a decision to loosen its existing standards. The initial results and the estimate of wider access are company-supplied.
The announcement does not give a sample size, approval-rate comparison or repayment outcomes for the new model. That leaves a material distinction between maintaining a lending decision threshold and demonstrating how the newly approved borrowers ultimately repay. Propel’s early approval finding does not itself provide that longer-term outcome.
This is not Propel’s first use of AI in lending. The company says it introduced its first proprietary AI model in 2015. Since then, Propel and its bank partners have underwritten more than two million loans and lines of credit, building a proprietary dataset focused on consumers underserved by traditional financial institutions.
Beyond approvals for its existing products, Propel says the cash-flow capability creates opportunities for more personalized lending and a broader range of financial products over time. Those are potential uses, rather than additional products launched with this announcement.
The next concrete rollout milestone is geographic: Propel expects to bring the model to the United States in early 2027. That remains an expansion plan, not current U.S. availability.
Loading discussion...
Join the conversation
Explain what would make that exchange worthwhile—or not.
Be the first to share a perspective or an experience.
Reader comments
Newest comments first. Replies stay oldest first.