Overhyped, but ignore at your peril: AI in Wealth Management and Private Banking

Another day, another perspective on Artificial Intelligence (AI). There is a continuous stream of announcements of AI being deployed across the B2C investments value chain – all are progressive but all are discrete internal processes / functions with a human still controlling the final output. This isn’t changing the world, but it will eke out a few basis points efficiency, similar to how the digital trend eked out incremental benefits but not brought revolution.

This pace is expected for the highly regulated, highly personalized Wealth and Private Banking industry. As one Global COO said, “the technology sector’s raison d’être is disruption; our industry’s is fiduciary”. Firms must experiment with ambition, with caution and within a robust set of tolerances due to the three commonly stated risks of:

• Damage to very personal client relationships;
• Novelty of governance and control around AI;
• Legislative and regulatory concerns.

However, these risks can also provide excuses not to address some of the strategic changes that can be made in the meantime.

Most operating models in our industry are far from being able to support AI at scale across the organization, and herein lies the threat of AI to the Wealth Management and Private Banking industry. Our market economics are going to change fast:

• Experts predict that AI will add ~$12trn to the global economy over the next decade. Of this, 60% (or ~$7trn) is expected to come from efficiency savings and the other 40% from new revenue [1];
• Both of these indicate a market operating on thinner margins. Researchers suggest productivity savings of between 15-60% from AI; even at the lower end of this, which is where Asset Management has been conservatively been suggested to sit, the potential for .competition is material [2];
• AI is (understandably) being grouped in a category of life-changing inventions such as the internet, the mobile phone and the automobile. Adoption trends have accelerated from 100 years with the car to 30 years with the internet to reach “peak” adoption. AI is expected to reach peak adoption by 2030 [3].

If AI follows this pattern, taken with other trends in the market (regulation, economic, stock market efficiency / pursuit of alpha), the existential threat to Wealth Managers and Private Banks (buy or be bought) is only going to gather pace – and firms in this sector do not change fast.

By 2029, based on Evercore’s projections, AI will be commonplace in the global population. This is credible; there are 8bn mobile phone subscriptions on the planet, giving most people access to ChatGPT at the tap of a screen. We can expect clients to be exerting competitive pressures on their providers by or soon after 2029.

All is not lost. Indeed, the forward-thinking leaders in our industry are experimenting hard and seeking out ways to win, rather than shying away from these disruptive forces. Our guidance to Executives is not to fixate on AI today but to prepare for an AI-dominated world in 5 years. Practical plans, reliable delivery and decisiveness will make winners stand out. Here are 5 initiatives firms should be considering:

• Ramp up experimentation. Existing operating models may limit this, so creative thinking will be required, including greenfield approaches. Learn from and build on others; don’t try what your providers have proved with others – now it’s been proved, it will be easier in future. Mix easy and difficult use cases, in particular how to deploy AI in the Front Office and explore how AI could reshape service propositions.
• Remove paper as mandatory. Functionality gaps, particularly in the Front and Middle Office, and compliance policies are generating processing weaknesses, risk, data issues and need for manual intervention – all are blockers to AI adoption.
• Reorganize operating activities and model for thinner margins. Many models, particularly insourced models, operate on a fixed ratio of new business to additional cost with limited flexibility to lower the unit cost with scale. Firms need to define commodity activities (establish how to conduct at scale) versus those that differentiate (retain control).
• Establish a data capability as the kernel of the organization. Data is a critically important asset and integral to proliferation of AI and its benefits across the organization. This needs new tools, new skills, new functions – the Target Operating Model which historically has had a Core Platform at its center will have Data at its center.
• Re-think risk management. The role of people will, as AI takes hold, reduce down to oversight and control functions and provision of emotional and unstructured service in place of processing. Processes such as meeting notes, attitude to risk, annual reviews / reporting, asset allocation, payments can all be robotized. The transition to AI enablement is risky, could be costly and will need different skills in the workforce.

[1] Professor David Shrier at FT Live PWM Conference, May 2024 and https://davidshrier.com/keynotes

[2] Google results “How much can AI improve productivity in asset management?, May 2024

[3] Evercore ISI, August 2023