Remaking, not replacing, the broker

One could make the case that mortgage broking is one of the most AI-proof professions out there.

Sure, AI is changing the face of mortgage broking before our very eyes, but human-centricity is baked into the value proposition. A mortgage is, for most people, the single largest financial commitment they will ever make – a decision no calculator can weigh on its own. When the stakes are that high, people want to deal with people, not a machine. The tepid uptake of digital mortgages validates this thesis.

Talk to the people pushing broker tech forward, though, and a clear theme emerges: it’s making busy work easier for brokers. It may sound trivial, but when half of broking has historically been chasing documents, re-keying data and cross-checking figures against lender policy, the impact is immense.

“Technology delivers its greatest value when it enhances, rather than replaces, the broker–client relationship,” says Leo Gagic (pictured, left), chief executive of the Finance Brokers Association of Australia (FBAA).

“By reducing costs, automating routine tasks, and improving efficiency, AI and associated technology allows brokers to focus on what clients value most: trust, guidance, communication and long-term relationships.”

Nicole Devine (pictured, centre), group executive general manager, product financial services within REA Group-owned Mortgage Choice, describes the same design principle. Mortgage Choice’s AI roadmap, she says, was built to answer one question: “How can AI cut back on the most time-consuming, repetitive tasks in a broker’s workload, which can be spent on the parts of broking that require human experience and judgement?”

Russell Graham (pictured, right), head of product and design at AFG, strikes a similar tone: “We started from a simple question – which parts of a broker’s day eat time without adding value for the client?” No prizes for the right answer: document handling and the sheer complexity of keeping up with credit policy across dozens of lenders.

It is evident that broker tech is evolving with these circumstances in mind.

Inside the toolkits

For an organisation representing 14,000-plus brokers, the FBAA sees its role as steering adoption rather than building its own products.

“Our past research has shown us that older brokers are less inclined to adopt new technology and AI, but this can change as we educate them more and they see the benefits,” Gagic says, pointing to events like the FBAA National Industry Conference as a venue for brokers to compare notes on what’s working and what available options best fit their business needs.

Read more: Should Australia brace for an AI-powered housing shortage?

The centrepiece of Mortgage Choice’s toolkit is Gemini Gems, a suite of more than a dozen custom AI assistants covering everything from marketing and compliance to document verification.

“They sit within the Google Workspace environment that brokers already use, which reduces the need to learn a new platform and helps drive adoption,” Devine says, adding that each Gem was “built around the problems [brokers] have told us are consuming their time,” with brokers directly involved in the design.

The clearest productivity gain to date has come from compliance. “In one of our largest businesses, pre-submission compliance review previously took one to two hours per file,” she says. “Following the introduction of the Compliance Review Gem, that process was reduced to approximately 10–20 minutes per file, freeing a full-time staff member to focus on packaging and processing work that supports settlements.”

An AI Academy of eLearning, webinars and training has also been “a critical factor in the success of our AI solutions”.

AFG’s toolkit targets two pressure points: documentation and policy complexity. Its AI Document Intelligence tool “classifies and organises files as they arrive, runs completeness checks and converts documents into pre-filled fact-find fields,” Graham explains. The second, Lender Insights, “puts the right guidance in front of the broker at the point of submission, which means fewer avoidable declines and less rework,” drawing on data from “more than 4,300 brokers rather than one book of business”.

What’s driven adoption in Graham’s view is the clarity around using the tools responsibly – brokers want guidance on where the line sits on client data and disclosure. “Brokers value AI most where it removes administrative load rather than where it touches customer guidance. They want their judgement to remain the ‘product’.”

As a result, AFG’s BrokerEngine Plus has averaged an NPS of +41 this financial year.

Tackling bottlenecks

While time to yes is an important metric for brokers, deal flow bottlenecks tend to emerge in other areas.

“The biggest challenges remain document verification, varying lender processes and manual assessment requirements,” Gagic says. “Technology is helping overcome these bottlenecks through automation, better system integration and improved data sharing.”

Read more: What APRA's AI warning means for your brokerage

Graham puts the friction at the pre-submission stage. “Incomplete files, documents chased in sequence rather than up front and policy mismatches that only surface once an application is with the lender slow the process.” This is exactly the gap AFG’s Document Intelligence and Lender Insights are built to close.

Mortgage Choice has made huge advancements in tackling annoying bottlenecks. A compliance review that once took one to two hours per file now takes as little as 10 minutes with its Compliance Review Gem.

Bottlenecks are also being improved when managing referral networks.

“Technology is helping brokers manage referral networks more effectively through improved communication, tracking and client relationship management,” Gagic says, pointing broadly to CRM systems and marketing automation as the categories doing this work across the industry.

Devine points to a specific example built into Mortgage Choice’s CRM: intent insights drawn from realestate.com.au and PropTrack data, which “helps brokers identify whether a client may be indicating buying or selling intent”.

AFG’s Partner Connect tool, built into BrokerEngine Plus, lets a broker “raise a referral from the deal they are already working on, with the client’s details carried across rather than re-entered, and then track its progress,” meaning “brokers capture opportunities they previously let go, without adding to their workload”.

Humans have the final say

Devine describes a division of labour inside Mortgage Choice’s toolkit that is representative of how the broader industry is evolving. “AI can help surface inconsistencies, missing information or documents that do not align with the expected requirements, but brokers are ultimately responsible for their judgement and decision making.” It’s the same principle she applied earlier to the Gems generally – support the work around a decision, never replace the decision itself.

Graham sees that same division of labour only deepening. “AI handles the analysis and surfaces the recommendation, and the broker makes the call.”

Gagic adds, “Technology is becoming increasingly important in fraud mitigation and prevention through timely identity verification, document authentication, data analytics and AI-driven machine learning risk detection.”

He frames the payoff less as a compliance box-tick and more as a trust dividend across the whole chain. “This helps protect clients, support lender confidence, preserve brand integrity and ultimately ensure legitimate borrowers can move through the process more smoothly and securely.”

In the end, the whole argument for why the broking profession might be uniquely AI-proof runs through the philosophy of these innovations. As Gagic says, “The real value is that these tools create more time for meaningful human interaction rather than replacing it.”