How the world’s largest banks build AI into the credit process: five examples

AI in credit has moved past pilots at the largest banks. An industry tracker updated in October 2026 counts 44 banks with AI in at least one lending step in production. Here are five examples of how leaders use it, based on public reports.

1. Wells Fargo: underwriting agents trained on past decisions

Evident cites Wells Fargo as a case study: AI agents trained on years of past loan decisions help underwriters review applications, while the human underwriter keeps the final say.

2. BMO: a commercial credit agent

BMO has built an AI agent for commercial credit that helps bankers prepare and review business loan cases, one of several generative and agentic projects banks launched in 2025–26.

3. DBS: agents that assemble credit memos

DBS uses agents to put together corporate credit memos, pulling financial statements, exposures and notes into the draft a credit officer reviews.

4. Synchrony: cash-flow underwriting

Synchrony is among the banks running AI in production in lending, with an approach based on cash-flow underwriting: decisions draw on how money moves through a customer’s accounts, not only on a credit score.

5. JPMorgan: AI across the bank

JPMorganChase topped the Evident AI Index for the fifth year in a row in October 2026, ahead of Capital One and RBC. The index measures how far banks have built AI talent, platforms and use cases across the business, including credit.

The common recipe

• Start with documents: spreading financial statements and reading bank statements saves the most time.
• Let AI draft, let people decide: memos, summaries and proposed decisions go to an underwriter.
• Use more data than the bureau: cash flow and transaction data explain thin files and small businesses.
• Keep reasons for every decision, so declines can be explained to customers and regulators.
• Govern it: owners, logs and monitoring for every model and agent.

Mid-size lenders can follow the same recipe without a team of hundreds. lendhero’s underwriter and document reader agents draft decisions with reasons and read statements, and the decision engine combines bureau, bank and cash-flow data, with the lender’s team keeping the final word.

Sources: Evident AI Index, BankingNewsAI, Quartz.