Insurance validation and vendor due diligence for QuickFi

QuickFi builds an embedded lending platform for business equipment finance. Equipment makers and banks use it to offer loans at the point where the equipment is sold. When we met QuickFi in early 2025, two routine tasks were taking up their operations team's time, and the work grew as the business grew.

The first was insurance. Every borrower has to keep commercial insurance on the equipment, and someone had to confirm it. That meant emailing the insurer, waiting for a certificate, opening the PDF, checking it against the contract, updating the record, and following up on anything missing.

The second was vendors. Before QuickFi pays for a piece of equipment, it needs to know the seller is a real, registered business that isn't on a sanctions list or in the news for fraud. That research was slow and done differently every time. We built an agent for each task.

Insurance certificates

Different insurers sent different certificate layouts, so the first extraction attempt was not always reliable. The reader needed fallbacks, while the surrounding workflow had to handle missing attachments and out-of-order replies without losing track of the borrower. We built that workflow on Flask and Supabase.

Vendor checks

The vendor agent runs independent checks and combines the results into a risk report for manual review. The checks include conflict-of-interest detection, such as whether a borrower owns the equipment vendor.

Secretary of State registries varied from state to state and changed without warning, so each needed a separate scraper and ongoing maintenance. Entity resolution meant handling names such as "Acme Equipment Inc." in one source and "ACME EQUIPMENT, INCORPORATED" in another. We used fuzzy matching to decide when two records referred to the same company. Google and Dun & Bradstreet rate limits also required request batching and retries.

Who built what

• Merek Sorianoled the insurance project
• Erik Pedersenbuilt the certificate validator, which classifies certificates and checks them against QuickFi's standards, and set up storage and the code connecting email, validation and the backend
• Brandon Xubuilt the API endpoints for borrower lookup, certificate upload and expiry dates
• Alex Kozikbuilt the data models and the templated request and follow-up emails
• Nicole Luobuilt the inbox monitor that matches incoming certificates to borrowers
• David Leeled the vendor project
• Audrey Lewellenbuilt the Google Maps and Search checks and the adverse-news pipeline
• Morgan Stuartbuilt the Dun & Bradstreet integration, conflict-of-interest detection and the risk scoring
• Zach Riiffbuilt the backend for vendor records and the external checks
• Arshdeep Kaurbuilt the APIs that talk to outside data sources
• Jonathan Zhangbuilt the Secretary of State and OFAC lookups

Spring 2026

QuickFi asked us back, and Mateo, Brandon and Eddie continued extending the vendor pipeline.

• Read the original case study (PDF)↗