Internal adviser toolkit

YMT Workbench

A set of AI tools for the Your Mortgage Team advisers — pre-check applications against lender policy, match clients to the panel, and answer policy questions, all with a citation back to the source for every claim. Built this week on public lender information we collected ourselves, with your lender panel and your towns in it.

Open Application Check View the pipeline

✓Cited to the source✓Runs offline (grounded fallback)✓Built on public data✓Transparent AI cost

What’s inside

7 lenders · 5 demo applications

✓Core

Application Check

Drop in a client’s application pack, pick a lender, and get a cited READY / NOT-READY pre-check before it goes to the bank.

→Checks LVR vs the lender’s caps and the RBNZ speed limit, deposit & gift-letter rules, income treatment, DTI and missing documents
→Claude writes the summary; a second AI pass strips anything it can’t cite
→Re-run against a second-tier lender to see the verdict change

Open Application Check →

⇄Core

Lender Match

Score a client across the whole panel — hard exclusions with cited reasons, then a ranked shortlist.

→Deterministic scoring — fully auditable, no AI in the ranking
→“Why it fits” and “watch out for”, each with a policy citation

Open Lender Match →

💬Core

Ask AI

Ask any lender-policy question in plain English and get an adviser-to-adviser answer with citations.

→Searches the whole policy library and answers comparatively
→Every specific claim links back to the public source

Open Ask AI →

☑Core

Fact Find Check

Turn a client’s situation into a precise, per-applicant document request — nothing missing, nothing unnecessary.

→Driven by an encoded NZ document-requirements rulebook
→Produces a ready-to-send (never auto-sent) client email

Open Fact Find Check →

📚Reference

Policy Library

The grounding corpus — the public lender and regulator sources every answer and finding cites back to.

→Browse by lender
→Every citation in the app links straight here

Open Policy Library →

◷Reference

AI Usage & Cost

Every AI call, metered at list pricing — transparent economics, no surprises.

→Per-tool totals
→See exactly what a check or an answer costs to run

Open AI Usage & Cost →

How it works

A rules engine does the maths

LVR, deposit, DTI, document completeness — all computed and checked in code against each lender’s criteria. Deterministic and auditable.

Claude explains it

The AI turns the findings into a clear, adviser-to-adviser summary or answer — it never invents the numbers.

Everything is cited & checked

Each claim links to a public source, and a second AI pass removes anything it can’t back up. It all still works with no API key.

Demo runs on public lender information we collected ourselves (accessed June 2026). The production build plugs into YMT’s own Trail fact-finds and aggregator policy library.