Loan Interviewers And Clerks

Blend Labs (NYSE: BLND) is deployed at over 300 financial institutions including JPMorgan Chase, Wells Fargo, and U.S. Bank, processing millions of mortgage and consumer loan applications annually with an entirely digital, AI-assisted pre-underwriting workflow that requires no loan interviewer involvement for standard applications. ICE Mortgage Technology's Encompass AI, following ICE's $11.7B acquisition of Black Knight, now dominates the LOS market with integrated automation across origination, processing, and compliance — its AIQ product uses machine learning to read and index documents, automatically satisfying conditions without human stacking. nCino, deployed at over 1,800 financial institutions globally, automates commercial and consumer loan processing through its Bank Operating System platform.

The accuracy and generalizability of AI document understanding has crossed a critical threshold in 2023-2025. Ocrolus, which processes over $1 trillion in annual loan volume, reports 99%+ data extraction accuracy on bank statements and pay stubs using a hybrid human-in-the-loop and ML model that has now reduced human review to edge cases. Inscribe's fraud detection platform uses document metadata analysis and ML to detect altered documents with accuracy exceeding manual human review. More critically, GPT-4 class LLMs can now be prompted to extract and cross-validate information from unstructured tax returns (1040s, Schedule C, Schedule E) with high accuracy — a task that previously required trained humans with 1-2 years of experience to do reliably. Fannie Mae's acceptance of third-party income verification data (Equifax's The Work Number, Finicity's Verification of Assets) bypasses document collection entirely for a growing share of applicants.