Skip to content

Origination

AI loan origination in Australia: a practical guide for lenders

By Saby Saxena5 min read

Summarise with

In short

Where AI agents fit between enquiry and credit, what must stay with people under the NCCP Act, and how Australian lenders can start without adding risk.

Talk to almost any credit manager in Australia and you’ll hear the same story. Applications are up, files are messier than ever, and the people best placed to assess risk are spending their days chasing payslips and re-keying data. It’s not a shortage of talent. It’s a process that hasn’t kept pace with how borrowers now apply.

That’s why AI loan origination has moved from conference slides to genuine board conversations. Lenders, from the majors to non-banks and mutuals, are asking a sensible question: can AI agents take on the preparation work in origination without weakening responsible lending or creating new risk?

This guide is our attempt at a straight answer. It covers what AI loan origination actually means, where it fits in an Australian mortgage process, what should always stay with people, and how to get started in a way your risk and compliance teams will be comfortable with.

What AI loan origination actually means

AI loan origination is the use of AI agents to support the work between a borrower’s first enquiry and a credit-ready application. The agents sit alongside your loan origination system rather than replacing it. They collect information, check documents, follow up on gaps and package the file so your assessors can spend their time on judgement, not admin.

The difference from earlier automation is flexibility. Rules-based workflows and online forms work well when every borrower behaves the same way. They don’t. A self-employed tradie with two ABNs, a couple buying with a family guarantee, or a borrower switching from a casual role to permanent part-time all need a slightly different path. AI agents can interpret plain English, read unstructured documents and adjust what they ask for based on what they’ve already seen.

Why origination is under so much pressure in Australia

A few local factors make origination harder here than it looks from the outside. Around three in four new home loans in Australia are now written through mortgage brokers, so lenders are processing files prepared to very different standards across hundreds of businesses. Refinancing activity has been high since rates rose from 2022, which means more applications competing for the same credit teams.

Serviceability has also become more involved. Lenders regulated by APRA assess loans with a serviceability buffer of three percentage points above the loan rate, and expense verification is more detailed than it was a decade ago. Every one of those checks depends on accurate, complete information arriving at the right time. When it doesn’t, files bounce back and forth and turnaround times stretch out.

Where AI agents fit in the origination process

Most of the value sits in the early and middle stages of an application, where the work is high volume and fairly predictable but still needs care.

  • Borrower intake: a guided conversation that captures employment, income, liabilities and goals, rather than a 40-field form many borrowers abandon halfway.
  • Document collection: requesting the right evidence for each borrower. That might be the two most recent payslips for a PAYG employee, or tax returns and Notices of Assessment for someone self-employed, depending on your policy.
  • Document checks: confirming documents are recent, legible, complete and match the details in the application.
  • Gap and inconsistency detection: flagging, for example, a salary on a payslip that doesn’t match bank statement credits, or a liability that appears in statements but not in the application.
  • File preparation: assembling a consistent, well-organised file with a short summary for the assessor.
  • Status updates: keeping borrowers and brokers informed so nobody has to ring up to ask where things are.

None of this is glamorous, but it’s where most of the time goes. Getting it right means assessors see cleaner files, brokers get fewer follow-up requests, and borrowers get answers sooner.

You can see how these steps come together on our AI loan origination page.

What should stay with people

Under the National Consumer Credit Protection Act 2009, lenders must make reasonable inquiries about a borrower’s requirements, objectives and financial situation, take reasonable steps to verify that information, and assess whether the loan is not unsuitable. ASIC’s Regulatory Guide 209 sets out how it expects those obligations to be met. None of that changes because AI is involved.

In practice, that means credit decisions, suitability assessments, policy exceptions and anything involving a borrower in vulnerable circumstances should stay with qualified people. AI can make those decisions better informed by putting complete, verified information in front of the right person. It shouldn’t be making the call on its own.

There’s also a privacy angle worth planning for now. Changes to the Privacy Act passed in late 2024 mean that, from 10 December 2026, organisations will need to explain in their privacy policies when they use automated decisions that could significantly affect individuals. Lenders using AI in origination should be clear about where automation is used and where a person decides.

How to get started without adding risk

The lenders getting the best results aren’t trying to automate everything at once. They start narrow, prove it works and expand from there. A sensible path looks something like this.

  • Pick one workflow where outcomes are easy to measure. Document collection is a common starting point.
  • Keep a human approval step on anything that affects a borrower’s application or outcome.
  • Make sure every action the AI takes is logged, timestamped and easy to review.
  • Integrate with your existing origination and core systems rather than running a parallel process.
  • Agree on success measures up front, such as time to credit-ready, files returned for more information and assessor time per file.
  • Involve risk, compliance and your broker partners early, not at the end.

If you’re an APRA-regulated lender, it’s worth bringing your CPS 230 operational risk team in from day one too. Any AI provider handling a critical part of origination may need to be managed as a material service provider.

What good looks like after six months

Done well, AI loan origination shouldn’t feel dramatic. Brokers notice that they’re asked for documents once, clearly. Assessors notice that files arrive complete with a short, accurate summary. Borrowers notice that they hear back faster and aren’t asked the same question twice. Your compliance team notices that every step is recorded. That quiet consistency is the real win.

Frequently asked questions

Does AI loan origination replace a loan origination system?

No. AI agents are designed to work alongside your existing origination system. They take on preparation and follow-up work, while your system of record and credit workflow stay the same.

Can AI make credit decisions in Australia?

Lenders remain responsible for meeting responsible lending obligations under the NCCP Act. The safer approach is for AI to prepare complete, verified files and for qualified people to assess and approve them.

How long does it take to see results?

It depends on your systems and starting point. Starting with one well-defined workflow, such as document collection, usually makes results easier to measure and trust.

If you’d like to see this on scenarios like yours, book a demo.

Saby Saxena

Founder, onebank

See it on your own scenarios

onebank runs the work in between. Your team approves every decision.

More articles

All posts