The Real Problem
It's 2pm on a Wednesday and you're staring at a kitchen table covered in documents. Your client — a couple buying their first home in Christchurch — has just handed you three months of payslips from two different employers, six months of bank statements from ASB and Westpac, a KiwiSaver statement, and a rental agreement.
You need to extract income figures, calculate UMI, cross-reference bank statement expenses against the CCCFA's affordability requirements, verify employment details, and manually key everything into your lender's application portal. This single file will take you three to four hours of pure data entry before you even start on the lending strategy.
This is the reality for most NZ mortgage brokers. 80% report significant administrative pressures, and brokers routinely spend up to 60% of their time on non-revenue generating tasks — data entry, document chasing, reformatting information from one system to another. NZFSG members have described it as "spending 70% of time on work that should take 20%."
The numbers tell the story: a broker working solo without efficient systems processes 8 to 10 files per month. With proper systems and support, that same broker can handle 18 to 25 files. The difference isn't skill or effort — it's administration eating the day alive.
That's 15 to 25 hours per week of recoverable time. Time that could be spent meeting new clients, building referral relationships, or simply going home before 8pm.
Why Existing Tools Don't Solve This
NZ brokers have access to solid CRM and workflow tools. Trail manages your pipeline, tracks commissions, and automates task reminders. MyCRM provides a structured application workflow. Advice Link offers a 6-step compliance process. FileInvite creates client-facing portals where borrowers upload their documents directly.
These tools solve document collection well. FileInvite means you're not chasing clients for their bank statements via email anymore. Trail means you know where each file sits in your pipeline.
But none of these tools solve document processing. When a client uploads six PDF bank statements and three payslips to FileInvite, those documents sit there waiting for a human to:
- Open each statement and manually extract income figures
- Calculate average overtime, bonuses, and variable income components
- Cross-reference expenses across multiple accounts
- Identify undisclosed liabilities (BNPL payments, personal loans, credit card limits)
- Key all of this into the lender's application form
- Double-check the numbers against the source documents
No existing NZ broker tool uses AI to read, extract, and verify document contents. The documents arrive digitally, but the processing is entirely manual.
How AI Solves This
You receive a notification that your clients have uploaded their documents. You forward them to your AI assistant via WhatsApp:
"New file — Sarah and James Cooper, first home buyers. Dual income. Documents attached. Please extract income, expenses, and flag anything unusual."
Within minutes, the AI returns a structured summary:
Cooper File — Document Extraction Summary
Sarah Cooper — Income Employer: Waikato DHB (permanent, full-time) Base salary: $78,500 p.a. Regular overtime: avg $485/fortnight (last 6 months) Estimated assessable income: $78,500 base + $12,610 overtime (80% of avg) = $91,110
James Cooper — Income Employer: Fletcher Building Ltd (permanent, full-time) Base salary: $92,000 p.a. Car allowance: $12,000 p.a. (non-assessable by most lenders)
Monthly Expenses (Combined) Rent: $2,400 | Groceries: $980 | Insurance: $320 | Utilities: $280 Subscriptions: $145 | Transport: $560
Flags
- Afterpay transactions: 3 active, totalling approx $680 outstanding
- Credit card (Westpac): $8,000 limit, avg balance $3,200
- Regular transfer to "J Cooper Savings" — $800/fortnight (likely KiwiSaver voluntary or personal savings)
Ready to pre-fill ANZ or ASB application. Which lender?
You review the summary against the source documents — a five-minute check instead of a three-hour extraction. You tell the AI to pre-fill the ANZ application template, adjust the overtime calculation to match ANZ's specific policy, and note the Afterpay balances for disclosure.
The Result
- 3+ hours saved per file on document extraction and data entry alone
- Fewer errors — the AI cross-references figures across documents, catching discrepancies humans miss when fatigued
- Undisclosed liabilities surfaced early — BNPL payments, secondary credit cards, and regular commitments are flagged before the lender finds them
- More files per month — recover 15+ hours weekly and reinvest in client-facing work
- CCCFA confidence — expense categorisation is thorough and documented, reducing the risk of affordability assessment challenges
What AI Can't Do Here
- AI won't make lending decisions — you still assess which lender suits the client and structure the deal
- AI won't guarantee 100% extraction accuracy — you must review the summary against source documents before submitting
- AI won't replace your relationship with bank BDMs — getting deals across the line still requires human advocacy
- AI can't access lender portals directly — it prepares the data, you submit it through the lender's system
Who This Is For
- Solo brokers processing 8-12 files per month who want to scale without hiring
- Broker groups where admin staff are stretched across too many advisers
- Any broker who spends their evenings doing data entry instead of being with their family
- Brokers who've lost clients because files took too long to process