Plaid Unveils AI Models to Enhance Credit Decisions Using Cash Flow Data

The company introduced a suite of AI-powered tools designed to help financial institutions make better decisions on credit risk, fraud prevention, and payment processing. Three new offerings centre on transforming how lenders assess borrower creditworthiness using live cash flow information rather than static credit scores alone.

Instant Link allows qualifying consumers to share their cash flow insights with lenders in seconds by connecting financial accounts to Plaid Consumer Reporting Agency. This gives lenders immediate visibility into repayment capacity, particularly for applicants who lack conventional credit histories.

LendScore 2 (Ls2) applies cash flow underwriting to predict loan repayment ability with 42% greater accuracy than traditional credit data alone. Plaid also released specialised versions of the model for auto financing, mortgages, and short-term lending.

LendScore Arc, a transformer-based risk model, analyses the sequence and timing of a borrower's transactions. Plaid describes Arc as its most advanced credit model to date, purpose-built for lenders ready to adopt transformer architecture in underwriting.

Michelle Young, Plaid's Credit Product Lead, explained that while transactional data offers a fuller view of borrower health, lenders have historically struggled to access and operationalise these insights at scale.

The next generation of LendScore and specialized models close that gap at scale, and with Arc, we're giving lenders new tools to expand access to more affordable credit, Young said.

The shift reflects broader industry movement toward real-time financial assessment. What a consumer earns and spends today increasingly outweighs backward-looking credit scores generated months earlier, as lenders seek to understand current repayment capacity rather than past behaviour.

Alongside the credit models, Plaid launched an AI foundation model trained on fraud patterns observed across its network. This technology strengthens Protect, the company's fraud detection solution, by identifying suspicious activity based on behaviours seen in millions of transactions.

A sequential foundation model now powers Plaid's ACH risk engine within Protect, improving predictions of payment failures and fraud exposure. The enhancement comes as regulatory pressure mounts: 94% of surveyed firms are not yet fully prepared for Nacha rules requiring all ACH participants to actively monitor for fraud, according to recent PYMNTS Intelligence research conducted with Plaid.

Will Robinson, Plaid's Chief Technology Officer, emphasised that the models draw strength from the platform's scale. They're built on foundation models already trained to understand how financial behaviour evolves over time across the entire Plaid Network.

They build on foundation models that already understand how financial behavior unfolds over time, across the Plaid Network, and that means better decisions, and better outcomes, for our customers and the millions of people who depend on those services to manage their own financial lives, Robinson said.

The approach leverages network effects: as more institutions connect to Plaid, the models gain access to richer datasets, refining their ability to spot creditworthy borrowers and flag fraudulent transactions with greater precision.