Back

AI Agents

Retail Mortgage Underwriting

Mortgage Document Completeness Agent

Final pack check; suggests missing pages

Context

Built for banking within the Retail mortgage underwriting stack, this agent runs in the final stretch—before sign-off—to ensure every required document, page, signature, and date is present and in-date for the specific product and jurisdiction. It’s applied after the core reviews (income, collateral, title, appraisal) so the file doesn’t stall at closing over preventable omissions.

What it does

The agent assembles the working file and evaluates it against your product/jurisdiction checklist (application, IDs/KYC, income evidence, self-employed schedules where applicable, bureau, purchase contract, title/registry, appraisal, EPC, insurance/pledge, disclosures/consents). It confirms document presence, version, freshness windows, signatures/initials, and page counts; detects missing or cropped pages, expired proofs, mismatched names/addresses, and template deviations; and generates precise ask-backs that specify exactly which item, page, or clause is missing and where it belongs. Output is a pass/flag status with short rationales and links to the exact pages used.

Core AI functions

Document detection and template recognition; OCR and field extraction for names, dates, addresses, policy numbers, assessor/licence IDs; rule evaluation against product/jurisdiction checklists and freshness thresholds; pagination and attachment integrity checks (missing pages, duplicates, wrong order); cross-document consistency (parties, property identifiers, amounts, dates); and reason-code generation with page-level lineage.

Problem solved

Missing, stale, or mis-packaged artefacts are often caught late, triggering rush ask-backs, re-underwrites, and closing delays. Manual file checks are time-consuming and inconsistent across teams.

Business impact

Files reach clear-to-close faster and cleaner: last-minute surprises drop, rework shrinks, conditions clear sooner, and audit posture improves because every checklist item has evidence and lineage.

Integration and adjacent use cases

Integration complexity: Low. The agent ingests documents from the borrower/broker portal, email, or DMS and writes the checklist outcome, flags, and targeted requests back into the underwriting workbench—no core changes required.

It is commonly combined with:

  • Income & employment validator,

  • Self-employed income validator (to ensure the right income artefacts are present),

  • Debt & liability disclosure checker (for bureau/statement evidence),

  • Purchase agreement & contract validator,

  • Land registry & title deed validator,

  • Appraisal report consistency,

  • EPC verifier, and

  • Collateral documentation verifier so completeness is verified after each specialized check and before final approval.

Bucharest

Charles de Gaulle Plaza, Piata Charles de Gaulle 15 9th floor, 011857 Bucharest, Romania

San Mateo

352 Sharon Park Drive #414 Menlo Park San Mateo, CA 94025

© 2025 FlowX.AI Business Systems