The editor considers emerging fields such as "agentic AI" and how fast, or not, wealth managers, private banks and others in the sector should embrace it, and how the benefits of using AI seem to vary widely.

More evidence is coming in that wealth advisors, bankers and others in our industry are using AI as part of their jobs. Almost every day, I receive email releases and comments about how artificial intelligence applications are getting a “seat at the desk,” as it were, and how firms are testing out ways to employ this technology. 

For example, late last week in a US private banks and trust companies report, Cerulli Associates said that advisors and home-office executives at banks have increased their use of AI for assisting with data reviewing and directly in portfolio construction and asset allocation. While just fewer than half (42 per cent) of bank advisors report using AI capabilities within their practice, this number is expected to soar to more than three-quarters (77 per cent) within the next two years. Private banks, cite “significantly higher” levels of AI usage with more than half (56 per cent) already using AI assistance to some degree and 80 per cent anticipating integration in the next two years.  

The range of stories about AI in this publication in the past few weeks shows just how busy this sector now is: 

Canoe Intelligence, a financial technology company powering alternative investment intelligence, has launched Canoe Labs, an incubator which allows investment and operations professionals to bring new AI capabilities to life. 

Broadridge Financial Solutions, a global fintech, has taken a minority stake in Uptiq, an AI platform for financial services. 

Advisor CRM, a client relationship management platform for RIAs, has unveiled its AI Meeting Assistant.

Envestnet, the US-headquartered turnkey asset management program and provider of fintech-driven back office services, has unveiled two AI innovations: Generative Business Intelligence (Gen BI) and Insights AI. The offerings are designed to transform the way advisors access, interpret, and use data.

This just scratches the surface of this news service's reportage on A1. If we wrote on no other topic, there would still be plenty of content.

Variations
While the overall trend appears to be towards more AI adoption, innovation and product/service rollout, there are differences to watch. A theme that comes up is whether firms can afford to be at the front of the pack in spending money on tech that might be outmoded within a year or less, whether they should try to be in that “middle space” or end up as a laggard. On 26 June, Bloomberg Professional Services said its late-2024 survey highlighted a growing divide between early AI adopters and laggards: nearly half of banks expect lower costs in the next three to five years (half predict a 5 to 10 per cent fall), while more than 40 per cent predict rising costs.

It is easy to see why firms vary in their approaches. These technologies can be expensive. According to Future Processing, (27 March 2024), a European technology consultancy, AI project costs are influenced by a variety of factors including development, hardware, data quality, feature complexity, and integration with existing systems, leading to costs that can range from $5,000 for simple models to over $500,000 for complex solutions. It is easy to see why smaller private banks, for example, might prefer to outsource as much of this sort of work as possible – the same will apply to family offices, to give another example. Building solutions in-house is largely a matter for bulge-bracket banks. 

And while figuring out the in-house vs outsourcing calculation, managers must also keep up with rapidly changing jargon and terminology (one wonders whether those of a geekier persuasion who read lots of science fiction have a distinct workplace edge these days). There are “co-pilots” and “virtual assistants.” A relatively newbie is “agentic AI” (a term that means, according to an AI-driven search that I used to find out about it, “autonomous artificial intelligence systems capable of setting goals, planning, and executing complex tasks with little to no human intervention”).

The aforementioned Bloomberg report said agentic AI is going to be major force: it can “handle complex workflows like resolving customer queries, optimising account balances and executing transactions.” But reaching the chosen destination will not be quick, because it requires large technology upgrades and making these new systems fit with core platforms. The report said, “data governance, legacy systems and regulatory scrutiny suggest the path to full autonomy could take more than five years.”

Given the dramatic space of change – look how systems such as ChatGPT have caught on – five years is a long time, although the sort of timeframe, it should be said, that people will often hold a private equity investment for. 

It must be hard for regulators to keep pace with this. While there may be clear benefits to developments such as agentic AI, it is easy to see how this will also make watchdogs nervous. In the UK, for example, UK government guidance (5 June) posed the risk of complete autonomous AI in cutting out human supervision – the stuff of regulators' nightmares and maybe also those of clients – and rolling out such technology prematurely. That takes us back to the question of how fast to roll out a technology – should one try to be at the head of the field and capture first-mover advantage or somewhere in the middle to avoid being burned by a leap too far? That appears to be a very difficult question to answer.