Human-enabled AI shapes the future of wealth management

Artificial intelligence (AI) has moved from the periphery of private banking to become a core part of banks’ strategies, driving efficiency, process improvements and business growth. Machine learning has been used in banking for many years, but large language models have triggered a step change over the past couple of years, making information that was once restricted or expensive more widely accessible.

Wealth management clients are increasingly adopting AI in their daily lives, including for financial matters. To understand this trend, HSBC partnered with Ipsos on its inaugural Global Affluent Report, surveying 9,993 affluent and high-net-worth (HNW) investors aged 21 to 69 across 10 global markets about their use of AI.

Ishan Sarkar, Head of Wealth and Premier Solutions at HSBC Singapore explained that the purpose of the study was to understand investor’s behaviour towards AI, including the value they place on AI in financial decision-making.

He said, “In addition, we also wanted to understand client’s expectations of the role of the relationship manager (RM) in the AI world. By understanding this, we can equip our RMs with the right tools and skill sets to serve the client of today.”

The AI trust gap

AI use among investors for financial analysis and research is very high, at 73%. However, only 12% globally and 8% in Singapore are willing to allow AI to make final capital-allocation decisions. This finding clearly highlights the AI trust threshold. While investors rely heavily on AI for research, they continue to turn to human advisers for judgement and final investment decisions.

Although AI use is high across generations, it is, unsurprisingly, highest among Gen Z investors. The survey highlights a profound cultural divide in AI’s influence on investor psychology and risk-taking. Investors in developed markets such as Singapore and the UK use AI strictly as an efficiency tool, optimising data gathering without altering their underlying risk parameters. In emerging markets such as India and Mexico, the survey found that AI acted as a psychological and operational catalyst, increasing investors’ willingness to take calculated risks.

Sarkar explained the trust gap using Singapore as an example. According to the survey, 79% of Singaporean investors said they needed reassurance and human validation before making major financial decisions, while 71% cited the need for a human adviser’s strategic judgement to synthesise broad market conditions into their long-term plans.

Sarkar said, “AI is great at completing tasks rapidly, but there is nuance in making a financial decision. People understand that a human is better able to make that decision than AI. AI can give you a stock recommendation, but clients rely on RMs to explain how it relates to their life goals and overall portfolio, and to separate market signals from noise.”

Human-enabled AI and the 'decision fitness' framework

Investors are now adept at using AI to conduct their own research and develop a shortlist of ideas. In some cases, they also stress-test these ideas themselves, a task once reserved for professionals. The survey found that although AI leaves clients better prepared for investment discussions, they still rely on an RM’s advice when making the final decision. This has led to the concept of human-enabled, or human-empowered, AI.

Sarkar explained the concept using the suitability framework, which evaluates whether a product is appropriate for an investor based on the investor’s risk profile. Standard suitability frameworks operate largely through process-based matching, aligning a client’s stated risk profile with a product’s parameters. AI can improve the efficiency, speed and data precision of this matching process.

However, a significant element of judgement is required when several products satisfy the client’s baseline risk profile and regulatory requirements equally well. Choosing option A rather than options B, C or D requires human insight and discretion. This level of nuance is why sophisticated clients continue to rely on RMs.

For a long time, one of an adviser’s primary responsibilities was to provide information to clients. Today, AI tools can perform much of that function, shifting the adviser’s role from information provider to curator of judgement.

HSBC’s report outlines a framework called “decision fitness”, which aims to identify the areas in which investors can rely on technology and those in which human advice matters most. The framework covers six areas: endurance, agility, stretch, strength, balance and recovery.

Sarkar cited the example of May 2026, when South Korea’s KOSPI index fell 20% in one day, rose 12% the next and then declined 15% the following day. Despite the extreme daily volatility, the South Korean stock market moved by less than 5% over the week.

He explained that relying entirely on an AI tool under such circumstances could produce a very different recommendation from that of a human adviser. An adviser can provide emotional validation by distinguishing meaningful signals from market noise and explaining how the volatility can be managed.

He said,“We call this ‘agility’ under the framework. Another important aspect is contextualising decisions against long-term life goals, which we call ‘stretch’. This is how advisers provide emotional validation, and it is exactly what clients expect from RMs today.”

AI use cases in wealth management

Wealth management, like the rest of the banking industry, faces a continuous influx of information. For banks, data collection has not been the primary bottleneck; the challenge has been consolidating and analysing the data. Generative AI has fundamentally changed this equation.

Sarkar said, “We have always been rich in data, but it was fragmented and difficult to use. AI overcomes these roadblocks and efficiently analyses data across platforms. Data mining and analytics for hyperpersonalisation are the biggest AI use cases in the wealth management industry.”

Using the Wealth Intelligence platform, HSBC RMs can now query more than 10,000 internal documents, global research reports and Chief Investment Office (CIO) publications. Instead of spending hours searching through documents to understand institutional views on macroeconomic trends, an adviser can now obtain this information in seconds, even during a live client consultation.

Similarly, administrative burdens for staff are being reduced through productivity tools like "AI Prepared."In the past, an adviser’s capacity to meet clients was constrained by the hours required to compile comprehensive meeting packs and extract portfolio metrics manually. By automating data mining, these tools reduce hours of manual retrieval to less than a minute. This efficiency is helping advisers move away from standard information delivery towards more meaningful advisory conversations.

Implications for bankers in an AI world

The AI revolution has multiple implications for bankers, who need to be trained in AI and become comfortable using it in their daily work. Implementing these advanced AI tools also requires a massive cultural shift in staff training. Banks cannot simply deploy complex AI models and assume that staff will leverage them correctly.

Sarkar stated, "If I can draw the analogy, you can have the best F1 car, but if you don't train the driver to drive the F1 car, you will never win a race."

To solve this problem, specialised internal training facilities, such as HSBC's Wealth Academy, are training advisers to become skilled and comfortable in using AI systems. This training views AI not as a static software package with a definitive end-state, but as a dynamic tool that matures through human interaction.

Advisers regularly feed queries to the system, effectively training the models in production. The models in turn are able to provide more relevant market insights to advisers.

The roles of RMs and investment managers are also evolving, requiring employees to change their mindsets to meet new expectations. As RMs become AI-enabled, they have considerably more bandwidth for RMs to focus on clients rather than administrative tasks. Banks increasingly expect RMs to have deeper conversations and closer relationships with a wider set of clients than before.

This change coincides with evolving client expectations, as RMs are increasingly expected to provide judgement rather than information alone. The role of the investment specialist is similarly moving beyond explaining baseline financial products to RMs or clients. As AI handles those basic explanations, investment specialists will need to focus on creating customised structures and addressing more complex client needs.

A collaborative hybrid model in an agentic world

The next phase of financial technology is expected to involve agentic AI: highly autonomous systems capable of executing multilayered workflows, evaluating portfolios and identifying complex opportunities.

Banks are testing hundreds of agentic AI use cases across their organisations, some of which are expected to enter production over the next 6 to 12 months. Banks are also increasingly partnering with leading technology providers on these applications to balance innovation with robust controls.

Regulators expect banks to establish comprehensive governance frameworks and ensure that their AI models are explainable. The development of agentic AI is likely to alter client relationships over the next few years by allowing clients to compare wealth management platforms and fees more rapidly, intensifying competition across the industry.

By relying on AI to handle processing, data analysis, and hyper-personalisation, human wealth managers can focus on human-centric needs that quantitative models cannot replicate. These needs and skills are what clients in an AI world are demanding from RMs: clear judgement, authentic human connections and validation through volatile markets.

Ultimately, the institutions that dominate the future of wealth management will not be those that attempt full end-to-end automation, but those that understand the value of a collaborative human-AI hybrid model.