As AI innovations gather pace, wealth firms must learn to manage the expectations of private clients who believe their financial returns will be quickly boosted.

Wealth management has been passing through an era of digital transformation for the best part of a decade. The integration of artificial intelligence (AI) and machine learning (ML) has impacted the ‘journey’ of clients and expanded the toolbox of their advisers.

But after much hype and little action, new technologies are finally beginning to shape the way assets are being managed.

This change of focus is highlighted by Alex Kokolis, global head of wealth at MSCI, the index provider, acknowledging that the entire financial services industry “continues to explore ways to leverage AI across various operational processes”.

Wealth firms, however, have switched tack, to focus on “implementing scalable, rules-based frameworks that can offer algorithmic rebalancing recommendations”, he says.

“When these firms have the correct models or benchmarks for comparison and an algorithmic rebalancing platform in place, it can suggest portfolio adjustments that align with the firm’s house views and risk framework, while saving the team time and effort,” adds Mr Kokolis.

A key challenge which MSCI is addressing with its clients in the wealth space is the automation of total portfolio rebalancing across public and private positions, side-by-side. Nearly 40 per cent of wealth managers surveyed by the firm earlier this year said they were eager for a digital platform to provide a single interface to manage all assets within portfolios, according to MSCI’s Emerging Trends in Wealth Management report.

“Ultimately, it takes the right data and analytical models to compare mixed portfolio positions in this way,” Mr Kokolis explains. “As firms are increasing their allocations to private market investments, this will become a growing need across the industry, one that technology providers must be able to address.”

DeepSeek, a Chinese GenAI company whose family of models made waves in the AI industry earlier this year, emerged from an investment firm. This is no surprise, believes Tim Gordon, co-founder and partner at Best Practice AI, a company assisting businesses in using AI to build sustainable competitive advantages. The former hedge fund team had the “incentive” and “capital” to access both the talent and computer power required to build “cutting-edge” generative AI.

Portfolio management is just at the beginning of its AI journey, suggests Mr Gordon. “Tools that parse company results commentary to analyse CEO sentiment or drive ever more sophisticated and complex trading strategies are already critical competitive assets.”

However, there are risks with generative AI. According to a report by UK Finance and consultancy Accenture, generative AI models, such as large language models (LLMs), can produce biased, inaccurate, or inappropriate outputs. They have the propensity to “hallucinate,” according to Mr Gordon, and their “stochastic” workings remain a black box. “This lack of transparency and consistency means regulatory concerns create real limits to full automation,” he adds.

Struggling systems

AI systems struggle particularly in unprecedented market conditions, when the “rules of the game change,” where human intuition and experience remain crucial, explains Michael Mainelli, chairman of Z/Yen Group, a City of London think tank.

As AI becomes increasingly integrated into portfolio management, managing client expectations is another hurdle, as investors tend to “overestimate AI’s predictive capabilities or are uncomfortable with algorithm-driven decisions”. Additionally, firms need to ensure their AI systems comply with “evolving financial regulations” and maintain transparency to avoid legal scrutiny. The need for ethical AI use and the protection of sensitive financial data further complicates widespread adoption.

Other experts echo a similar sentiment. While AI may enhance certain areas of the investment process — for example, the search and primary classification of large amounts of investment data or certain areas of high-frequency trading — Miguel Burguet, managing director of Marlowe Capital, a macro investment manager focused on the US banking industry, is “quite sceptical” about reliability of AI systems throughout the entire investment cycle, particularly in times of extreme stress or turmoil, such as during a “Minsky Moment” or a 2008-style crisis.

“AI lacks strategic thinking and struggles with non-quantifiable factors like sudden political shifts or black swan events. You need a human for that… and they sometimes fail too,” he says.

Despite these complexities, AI is seen more as an enhancement than a replacement for human managers. “AI and automation are currently positioned as tools to enhance human decision-making rather than fully replace portfolio managers,” explains Mr Mainelli of Z/Yen Group.

While AI can process and analyse data more efficiently, human judgement remains critical for interpreting nuanced economic indicators, geopolitical events and investor sentiment. AI excels in quantitative analysis and high-speed trade execution, he says, but strategic decisions still rely on the qualitative insights and experience that human managers bring to the table.

That said, Mr Mainelli acknowledges that in areas like passive investing, algorithmic trading, and risk assessment, “AI-driven systems may eventually take on more autonomy, reducing the need for human intervention in repetitive or rules-based tasks.”