The Challenge
Manual document extraction was the bottleneck. Underwriters spent 1–2 hours per deal pulling numbers from bank statements, applications, and tax returns—copying values into spreadsheets, double-checking math, and hoping they didn’t miss a line item on page 14 of a 30-page statement.
Deal data lived across email inboxes, shared drives, Excel workbooks, and sticky notes. When a second underwriter touched a file, they had no idea what the first one had already done.
Stacking risk was invisible. Existing MCA positions were tracked in a spreadsheet tab that was always out of date. Lenders frequently discovered—after funding—that a merchant was already leveraged to 60%+ holdback across three other advances.
Offer turnaround killed deals. By the time an underwriter finished analysis, built a scenario, drafted an offer email, and got approval to send it, 2–3 days had passed—and the ISO had already placed the deal with a faster competitor.
The Solution
We built an AI-powered underwriting operating system:
- Multi-vendor extraction engine — Documents processed simultaneously by up to four extraction providers. Results appear side-by-side so the team can validate accuracy and pick the best output per document type—all within minutes of upload.
- Unified deal dashboard — Every deal has a single page with merchant info, financials, existing positions, documents, tasks, and a complete activity log. Real-time sync means two underwriters never overwrite each other’s work.
- Auto-computed position tracking — Enter three fields and the system calculates weekly pull, monthly pull, term remaining, payback amount, and holdback percentage. Positions auto-downgrade to “Paid” when balance hits zero.
- Stacking risk detection — Automatic alerts when a merchant has 2+ active positions with short remaining terms or high holdback percentages.
- Scenario builder with one-click offers — Build multiple funding scenarios, then generate a professional HTML email with PDF attachment sent from the underwriter’s own email account. Every offer is logged.
- Complete audit trail — Every field change, every email, every decision is logged with user, timestamp, and before/after values.
The Results
- 3x deal capacity — same 8-person team went from 50 to 150 deals/month
- Document processing: 60 minutes → 2 minutes with 95%+ accuracy
- Approval decisions in 2–4 hours — down from 2–3 business days
- Same-day offers sent within hours of submission
- 20–30% reduction in loss rates from early stacking detection
- Zero compliance gaps — full audit trail on every action
Who This Applies To
MCA and SMB lenders processing 50+ deals per month where underwriters spend more time on data entry than credit decisions. If your team is bottlenecked by manual document extraction and spreadsheet-driven workflows, this addresses that directly.