TrendAI — Delivery Portfolio

Step 01·Lead intake·A referral email becomes an application file with one click.

Source

Referral · email

From: priya.malhotra@anz.com.au · Subject: 'Borrower referral — Nguyen family'

Hi Mai,

Want to introduce you to John & Sarah Nguyen — they're looking at a place in Carlton North VIC 3054, asking $890k. They've saved a 20% deposit. Combined income around $215k.

John: Telstra (8 yrs), Sarah: Melbourne Uni (~18 mo). Existing: $18k cc, $9k novated lease. Contact: +61 412 778 902 · john.nguyen@gmail.com.

Lead form attached.

Priya Malhotra · ANZ

New application file · AI pre-fill

AI parsed

Applicant 1: John Nguyen
Applicant 2: Sarah Nguyen
Contact: +61 412 778 902 · john.nguyen@gmail.com
Loan purpose: Owner-occupied purchase
Property: Carlton North VIC 3054
Estimated value: $890,000
Loan amount: $720,000 (LVR 80.9%)
Income (HH gross): $215,000 p.a.
Existing liabilities: $18k credit card · $9k novated lease

AI · Parsed 22 fields from referral email body + attached PDF lead form. 3 fields flagged 'low confidence' (employer ABN, second-borrower DOB) — broker review prompted.

Who acts

Broker (Mai)

What happens

Mai forwards a referral email into the LAMS inbox. The system parses the email + any attached lead form, drafts a new application file with 22 fields pre-populated, and asks Mai to confirm 3 low-confidence ones.

Why it matters

The Excel-era equivalent was 10–15 minutes of re-keying borrower details per lead. At HFHL's volume (~40 new leads/week), that's a half-day of manual entry the brokers don't lose anymore.

AI moment

Email + PDF parsing. The system extracts names, contacts, loan purpose, property, income and liabilities — and self-reports confidence per field so the broker knows where to look.

Tie to design

Side-by-side: the original email on the left, the parsed fields on the right. Confidence pills surface what to double-check, not what's certain.

1 / 6

That's the loop — intake to settlement, with AI doing the document and pattern-recognition work so brokers can spend their time on the conversation. The system is live with 20+ HFHL staff today.