AI is coming for your back office. Some will adapt. Many will not.

It has been, by almost any measure, a golden era for Australian mortgage broking. In the June quarter of 2025, brokers wrote a record 77.6 per cent of all new home loans — up from fewer than 60 per cent just four years earlier, according to Mortgage and Finance Association of Australia data compiled by Cotality. In the September quarter, brokers settled $130.23 billion in new home loans, the highest value ever recorded for any single quarter. More than 22,000 brokers are now operating nationally, serving a market that has increasingly come to depend on independent advice to navigate an extraordinarily complex lending landscape.

And yet, at the very moment the industry is celebrating its growing dominance, a sweeping new analysis from the Brookings Institution in Washington and the National Bureau of Economic Research is raising urgent questions about the people who make that dominance possible: the loan processors, the credit assistants, the compliance coordinators, the parabrokers, and the administrative staff who sit behind every settled loan.

These workers, the research finds, are among the most exposed to artificial intelligence automation in the entire workforce. They are also among the least equipped to recover if that automation displaces them. For a broking industry that has built its success on personalised service and operational infrastructure, that combination should be deeply concerning.

“Brokers must embrace rapidly changing technology on an ongoing basis, and as they do this sensibly and even cautiously, they won’t be left behind,” said the FBAA’s Peter White

What the research actually says

The analysis, published in January 2026 by researchers Sam Manning and Tomás Aguirre of the Centre for the Governance of AI, along with Brookings senior fellow Mark Muro, takes a different approach to the question of AI and work than most studies have attempted.

Rather than cataloguing which jobs artificial intelligence can theoretically perform — a measure of exposure — the researchers asked a harder question: if a worker loses their job to AI, how likely are they to land on their feet? To answer it, they built what they call an "adaptive capacity index" across 356 occupations covering 95.9 per cent of the US workforce. The index draws on four factors: a worker's liquid savings, the transferability of their skills across roles, the density of local job opportunities, and their age.

The picture that emerges is sobering. Of the 37.1 million workers in the top quartile of AI exposure, about 26.5 million also have above-median adaptive capacity. Software developers, financial analysts, lawyers — yes, their jobs are highly exposed to AI, but they tend to have savings, professional networks, and broad skills that make transitions manageable. For 6.1 million workers, that is not the case. These are people whose jobs are both highly exposed and who score in the bottom quartile for adaptive capacity. They are typically older, have limited savings, possess skills that do not easily transfer to other roles, and live in smaller cities where alternative work is thin on the ground.

The occupations concentrated in that vulnerable group read like an organisational chart of a mid-sized mortgage aggregator: office clerks, secretaries and administrative assistants, receptionists, payroll and timekeeping clerks, data processing staff, and compliance support workers.

Australia's particular exposure

While the Brookings research draws on American data, the structural picture it describes maps directly onto Australia's broking industry — and several local findings make the risk even more acute here.

Jobs and Skills Australia, the federal government's workforce advisory body, has specifically identified administrative and clerical roles as among those most exposed to automation risk in the Australian labour market. Analysis by the Social Policy Group estimates that 43 per cent of Australia's administrative and support services workers could experience displacement by 2030 if current AI adoption rates continue. And CSIRO's Data61 has warned that Australia risks uneven AI adoption, with productivity gains concentrated among large firms while workers absorb the adjustment costs.

Closer to home, the Australian Finance Industry Association has found that the finance sector's adoption of generative AI is on track to double over the next three years, with mortgage applications among the areas most targeted for change. New AI platforms are already emerging specifically for the broking industry: Cynario, launched recently by former MoneyQuest chief executive Michael Richardson, uses AI to help brokers navigate lender mortgage policies. LoanOptions.ai's HAILO tool is automating parts of the application process. Loan Market Group has announced an expanded AI roadmap focused on broker efficiency. Non-bank lender Bluestone is using AI to detect fraud and accelerate application processing.

Each of these tools does something useful. Each of them also reduces the number of humans needed to perform the tasks those tools now handle.

LinkedIn's Jobs on the Rise 2026 report identified AI literacy as the most in-demand skill in Australia, with finance topping the list for AI-focused hiring at nearly 12 per cent of job advertisements requiring those skills. Put plainly, the industry is already shifting its hiring toward AI-capable workers. The logical corollary is that it is also shifting away from workers who are not.

An FBAA poll suggests a marked age-based difference in adoption. Brokers aged under 50 were twice as likely to use AI in their work (54%) as those aged over 50 (27%). The association linked this to the challenge some longer-serving brokers face in adjusting established practices to newer technologies.

The gender crisis hiding in the data

The Brookings research contains a finding that the Australian broking industry has not yet fully reckoned with. Of the 6.1 million workers in the US identified as facing both high AI exposure and low adaptive capacity, approximately 86 per cent are women.

That figure is not a coincidence. It reflects decades of occupational sorting that have concentrated women in precisely the administrative and clerical roles most susceptible to large language model automation.

For Australia specifically, the risk is even more pronounced than the global average. The Gender Snapshot 2025 report found that in high-income countries including Australia and New Zealand, 9.6 per cent of women's jobs are at high risk of automation, compared with just 3.5 per cent of men's. That is nearly three times the exposure — a gap that should give any principal or aggregator pause when reviewing their operational workforce.

The United Nations has found that women's jobs are more exposed to automation than men's, with clerical and administrative roles facing the highest risk. These are not niche roles — they are the doorway many women use to enter the workforce. In the broking context, that doorway typically leads to a parabroker or loan processing role: the entry point through which many industry participants first learn the business before advancing to senior positions.

There is a second, less-discussed dimension to the gender risk. Research by Harvard Business School has found that women are adopting AI tools at roughly 25 per cent lower rates than men. The reasons are multiple — ethical concerns about the technology, fear of being judged for relying on AI-generated output, and historically lower access to STEM pathways — but the consequence is straightforward: workers who do not build AI competency are at greater risk of being replaced by those who do, or by the tools themselves.

AI rarely eliminates entire occupations overnight. Instead, it erodes them task by task. Calendar management becomes automated. Report drafting becomes AI-assisted. For women in broking support roles, that gradual erosion may arrive before they have had the opportunity — or the support — to build the skills to respond to it.

The broker Is not off the hook

It would be a mistake for brokers themselves to read this research and conclude that the risk sits entirely with their support staff.

The tasks that define a significant portion of a broker's working week — gathering client documentation, comparing lender products, calculating repayment scenarios, preparing pre-approval letters, processing condition requests, following up on outstanding items — overlap substantially with what AI tools can already perform, and increasingly with what they are being specifically designed to perform in the Australian market.

The brokers who will be most resilient are not those who assume their client relationships make them immune. They are those who are actively embedding AI into their workflow and building the kind of demonstrated AI competency that the labour market is already rewarding. PwC Australia's AI leader Tom Pagram has noted that jobs requiring AI skills command a clear wage premium. That premium is a market signal — and it is already visible in the finance sector's hiring patterns.

The analogy to the Royal Commission is instructive. When the Hayne Royal Commission forced brokers to adapt to the Best Interests Duty, the industry feared disruption and emerged stronger. Those who adapted earliest, built compliance capacity, and made the new regulatory environment part of their value proposition gained ground. Those who waited lost it. The AI transition rewards the same posture.

What the industry should do

The Brookings research does not predict mass unemployment. It is a targeting tool — a way of identifying where the risk is concentrated before displacement arrives. For Australian mortgage brokers and their aggregators, several practical implications stand out.

Invest in upskilling support staff before it becomes urgent. The parabrokers and loan processing staff who are retrained to work alongside AI tools — managing complex exceptions, auditing AI outputs, handling the non-standard scenarios that automation consistently struggles with — are far more valuable than those who are simply made redundant. Principals who build that transition path now will retain institutional knowledge and avoid recruiting costs later. Aggregators that offer structured AI training as part of their value proposition will hold their networks.

Rethink the entry-level pipeline. Many of the roles most exposed to AI have historically served as the first rung on the ladder for new entrants to the industry — positions where people learn credit policy, lender requirements, and application mechanics before moving into senior roles. As those positions contract, the industry needs to think carefully about how the next generation of experienced brokers will be developed. The talent pipeline does not replenish automatically.

Brokers: build AI fluency, not just awareness. Knowing that AI tools exist is not the same as knowing how to use them to identify refinance opportunities in your trail book, prepare comparison documents more efficiently, or handle initial borrower enquiries outside business hours. The brokers writing the most loans in three years' time will be those who treated AI as a lever for productivity in 2025 and 2026, not as a threat to be monitored from a distance.

Take the gender dimension seriously. If your operational team is predominantly female — as it is in most Australian brokerages — you have a specific responsibility to ensure that AI upskilling is not left to chance or self-selection. Support staff who are not offered structured pathways into AI-augmented roles will not find those pathways on their own. The brokerages that design equitable transitions will build more loyal, more capable teams.

The deeper question

There is a particular irony in the current moment for Australian mortgage broking. The industry has spent a decade arguing — correctly — that human brokers provide something that banks and digital lenders cannot: personalised advice, genuine advocacy, and the kind of contextual judgement that cannot be replicated by an algorithm. That argument has won. More than three-quarters of all Australian home loans now flow through brokers, and the proportion keeps rising.

But the argument about human value applies unevenly within the industry itself. The broker relationship is human. The infrastructure that supports it — the processing, the compliance administration, the data management — is increasingly not. The people running that infrastructure are overwhelmingly women, often in smaller markets, often without the savings or the skill transferability to absorb a sudden job loss.

Australia's broking industry has earned its record market share through professionalism, advocacy, and genuine client service. The question now is whether it will apply the same ethic to the people inside its own businesses — and whether it will move before the displacement arrives, rather than after.