Case Studies

Growth-stage fintech: underwriting copilot cut memo time 61%

An SME-lending fintech was losing deals to slower underwriting cycles than its competitors. We shipped an underwriting copilot that assembles the credit memo, flags policy exceptions, and surfaces comparable deals — cutting analyst prep time by 61%.

Sector
Financial services — SME lending
Client size
~600-person fintech, Series C
Capability
Generative AI + Agentic AI
Time to production
10 weeks

The challenge

Underwriters were spending 3–5 hours per deal on memo prep: pulling financials, running comparable-deal searches, flagging policy exceptions, and drafting narrative sections. Deals were being lost to competitors with faster turnaround, and the pipeline was capped by underwriter capacity, not demand.

Our approach

Agentic Workflow Sprint over 10 weeks. The copilot integrates with the loan-origination system, retrieves comparable deals from the historical portfolio, pulls current financials via existing bureau connections, and drafts the credit memo in the approved format. It explicitly flags policy exceptions, cites its sources, and hands off to the underwriter for judgment and sign-off. Every model output is logged for model-risk review.

The outcome

Live in 10 weeks. Median memo-prep time dropped from 4h 05m to 1h 35m. Underwriter throughput up 43%. Approval rate on same-day decisions up 22%. Model-risk review completed with no material findings.

Memo-prep time reduction
61%
Underwriter throughput lift
+43%
Same-day decision rate lift
+22%
Time to production
10 weeks

Have a workflow like this in your business?

Most of the highest-leverage AI wins we ship are workflows the operator already knows are broken. Show us yours in a 30-minute scoping call.