Mortgage and lending (New Zealand)

What is in the early-access pack

The mortgage and lending pack is available to early-access customers. Document types cover loan applications, adviser fact finds and lender offers; payslips, employment agreements and commission statements; IRD income summaries, IR3 returns and accounts for the self-employed; bank, credit card, loan and KiwiSaver statements and gift letters; household expense declarations and bills; valuations, rates notices and insurance schedules; sale and purchase agreements, LIMs, records of title and unit title information; and KiwiSaver first home withdrawal documents, including the Kāinga Ora letter and a solicitor undertaking.

Starter schemas extract fields such as borrower, lender, loan purpose, requested loan, deposit, employer, gross pay and year-to-date figures, income summary totals, declared and observed monthly expenses, purchase price, valuation and settlement date. The extraction instructions ask for only the last four digits of the IRD number; Droqify does not currently mask these numbers. Cross-document indexes cover application reference, borrower, property, lender, employer, employment type, loan purpose, KiwiSaver provider, tax year and document kind. Smart folders group files by permanent employment, contract employment, self-employed income, purchase loans and refinance loans. Sample questions ask what income the documents show, what the declared and observed expenses are, and what is missing or needs human review.

What is still being confirmed

Early-access packs are built from public form and agency guidance; some details are still being checked and may change. They have not been validated on production customer documents, and forms and fields will be confirmed with early-access customers. Reviewers compare documents using the extracted data and search; the pack does not check a file for completeness or compare documents automatically. Where your forms are not covered, your team can define additional schemas.

Where people stay in the loop

Each extracted field has a model-reported confidence score (not calibrated). When you turn on Human in the Loop for an upload, processing pauses at the steps you choose so a person can check the source document and correct the data before it continues. Documents that match no schema are flagged. The pack prepares document data for your adviser or credit team; it does not assess affordability or make lending decisions. Key actions are written to an audit log.

Join the waitlist

New Zealand is in early access: not yet validated on New Zealand documents with customers. If you want to try the pack on your own documents, join the waitlist and tell us about your document mix and volumes. We will follow up directly.