Underwriting
How to use LLMs to summarise borrowers, draft credit memos, run underwriting workflows, and keep committee-readable audit trails. With prompts, scorecards, and honest failure modes.
AI credit scorecard vs ML model: which one actually fits a small-to-mid-sized lender?
The honest comparison between a rules-based AI-augmented scorecard and a trained ML default-prediction model — when to pick each, what they cost, and why the scorecard wins for most lenders under 10,000 active loans.
AI in loan collections: what actually works, what doesn't, and where compliance draws the line
A builder's view of where LLMs earn their seat in collections workflows — draft generation, segmentation, hardship conversations — and where they're a compliance liability.
Underwriting thin-file borrowers with LLMs: the diligence prompt chain that turns alternative signal into a credit decision
A four-step LLM diligence chain that converts gig earnings, rent ledgers, and telco data into the same scorecard columns a thick-file applicant fills out — for borrowers in any market.
Credit memo generation with LLMs: the prompt chain that survives a committee
A structured prompt chain that produces first-draft credit memos your credit committee will actually accept — with risk grade, conditions, deviations, and explicit gaps flagged.
How to underwrite loans with AI: a builder's guide (2026)
The full workflow for AI-assisted loan underwriting — from application intake to credit memo, with the prompts, scorecard logic, and failure modes that matter.