AI adoption in advice sector still stuck at pilot stage

Iress and Signal Advisory have published research showing that only 7% of advice practices have embedded AI across most of their workflows. The study found that firms with deeper AI use are up to three times more likely to report broader business benefits.

Drawing on responses from more than 800 advice professionals in Australia and New Zealand, the report suggests that capacity, rather than client demand, is a central constraint on growth. Some 81% of respondents said their business could absorb a 30% increase in adviser capacity.

This points to a sector where operational bottlenecks are a bigger issue than a shortage of work. The research found that AI is already delivering productivity gains, but in most firms those gains have not yet led to broader changes in business economics or the technology stack.

Across the survey, 72% said they were piloting AI or had moved beyond that stage, but only a small minority said the technology was embedded across most of the advice process. Practises that had integrated AI more deeply were more likely to say it helped advisers see more clients, improve service and replace existing software or external services.

Capacity limits

Advice-document production emerged as one of the clearest pressure points. Some 45% of respondents rated it as the heaviest part of the advice process, while 35% said they planned to use AI for statement of advice generation over the next 12 months.

Current use remains concentrated in narrower administrative tasks. The study found that 81% use AI for meeting notes and file notes, compared with just 8% for statement of advice generation.

Sam Wall, Chief Executive Officer - Wealth APAC, Iress, said: "AI is already delivering real productivity gains in advice. The opportunity now is to extend those gains further into the workflow and turn them into capacity, helping businesses serve more clients, improve the quality of the advice experience and grow.

"Advice document production remains one of the biggest burdens, yet AI adoption in this area is still low. Our ambition is to help advisers materially reduce the time it takes to produce advice, creating more time with clients and, ultimately, the potential to make advice accessible to more people.

"But achieving that depends on trust. While 89% of respondents use general-purpose AI assistants, 31% would not trust a large global AI company to deliver AI for regulated advice safely and effectively. Advisers need confidence in the data, technology and controls around AI.

This aligns with our work on the evolution of Xplan. AI and agentic capabilities are central to our product strategy as we make Xplan faster, modernise the client experience, improve workflows and bring AI into the advice journey. The goal is to make advice significantly more efficient while keeping advisers firmly in the driver's seat, using the trusted data, workflows and controls within Xplan as the foundation for AI."

Barriers remain

The findings also show why adoption has remained uneven. Half of respondents cited poor integration as a barrier to greater AI use, while 43% said they would prefer to wait for AI functions within software they already use rather than adopt a standalone tool.

Regulatory uncertainty is another factor, with 64% saying regulators have not provided enough clarity on the use of AI in advice.

Trust in outputs appears conditional rather than absolute. Almost half of respondents (48%) said they would accept AI-generated recommendations if a person reviewed everything before it reached the client.

The report suggests the sector is still working through where AI fits best in the advice process. While enthusiasm is widespread, many respondents appear to favour tightly controlled use cases with clear human oversight rather than full automation of regulated work.

Measuring impact

One of the more striking gaps in the research concerns measurement. While 58% said measured hours saved per adviser would be the strongest justification for further AI investment, only 28% said they had measured the time or cost impact of the tools they already use.

That leaves many firms unable to show whether apparent efficiency gains are being converted into lower operating costs, higher throughput or better use of adviser time. The report found that 47% of AI-engaged respondents said they save at least a quarter of their time on the tasks where they use AI, yet only 20% reported a material reduction in cost to serve.

Time saved in one part of the process can also be lost elsewhere. The study found that 30% of AI-engaged respondents said checking outputs consumed much of the time AI had saved.

Anthony Caneva, Founder, Signal Advisory, said: "The research shows a two-speed value story for AI. The productivity gains are real, but they are not yet consistently flowing through to the wider economics of the advice business. While 47% of AI-engaged respondents report saving at least a quarter of their time on the tasks where they use AI, only 20% report a material benefit in lower cost to serve.

"AI won't be the answer to every process challenge. Traditional rules-based automation remains well suited to deterministic, repeatable tasks, while AI is better suited to work involving natural language, interpretation and synthesis. The focus should be on applying the right technology to the right task.

"For advice businesses, that means looking beyond the AI task itself to what happens around it. Some 30% of AI-engaged respondents say checking outputs consumes much of the time AI saves. Hand-offs, rework and verification can absorb those gains before they translate into capacity or growth.

Measurement is also critical. While 58% say measured hours saved per adviser would most justify further AI investment, only 28% have measured the time or cost impact of their existing AI tools. Firms need to establish a baseline and track whether time saved is actually released back into the business."

The broader competitive pressure is already visible in sentiment across the sector. The survey found that 78% believe firms that have not adopted AI within two years will face a competitive disadvantage.

At the same time, the research suggests the next phase of adoption will depend less on experimentation and more on integration, measurement and trust. Just 5% of respondents said they had retired an existing software tool or service as a result of AI, while 63% said AI had been added to their technology stack without replacing anything.