Intelligent evolution: From machine learning to agentic AI in wealth management

I know what you are thinking. Another article about the opportunity of AI… You are ready to run for the hills, aren’t you? But humour me a minute, and I will try and convince you to stay.

Not only will I show you that that I care deeply about the changes that are happening in wealth management businesses, but hopefully I’ll also illustrate why I think it is so important and why you should care too. Right, let’s get started.

Financial services businesses are ledger businesses. Wealth management as a sub-segment of said ledger businesses are data, insight, knowledge and risk businesses. To grow wealth, you have to consistently and accurately predict trends, model patterns and bet on outcomes… all whilst managing the expectations and vulnerabilities of the people who give you their assets to manage. Not easy to do, and for most… impossible.

Now for a moment let’s think about data and how it’s used in these businesses.

Sometimes we read it, think about it and contextualise it, other times it merely reflects the objective facts…sometimes (if done correctly) it provides objectivity, subjectivity, nuance, subtext and more.

And if we’re really lucky, it does the heavy lifting in a way that a human or group of humans could never achieve. Enter Big Data, enter machine learning… enter AI.

For years, Rules-based automation worked to reduce manual processes, machine learning took us significant steps forward in understanding nuance, NLP assisted in the furthering of our aims gaining insight, AI took us closer to the promised land of computer based IQ… but the world is now hovering on the precipice of the biggest shift in a generation when it comes to the future of knowledge work.

This isn’t an ‘imagine if’ article, nor is it an ‘extinction is on the horizon’ article. It’s a pragmatic piece about how wealth management knowledge work might change in the coming five years and how firms might just do some of the most innovative work out there.

As Sarah Connor once said in the Terminator series “You taught us there is no fate but that which we make for ourselves.” So, no determinism here!

The AI crossroads

The wealth management sector stands at a pretty fascinating crossroads, where it is having to navigate the transformative power of artificial intelligence, whilst running their decades-old (in some cases centuries-old) businesses day to day.

What began as an exploration of basic automation and data processing has been rapidly evolving into a landscape where intelligent systems are poised to fundamentally reshape how wealth is traded, accumulated, managed and potentially transferred.

Charting the progression from the established applications of machine learning, all the way through to the anticipated impact of agentic AI fills some with dread, but for us, it’s incredibly exciting!

Machine learning has already become an increasingly influential force within wealth management, primarily serving as a powerful tool for enhancing operational efficiency and generating data-driven insights.

Its impact is evident across various critical functions. For instance, machine learning algorithms (ML) excel at analysing vast datasets to identify subtle patterns and construct predictive models, enabling wealth managers to better anticipate future market trends and asset price fluctuations. This capability allows for more informed decision-making regarding investment strategies, directly impacting the core value proposition of the industry: the ability to generate optimal returns for clients.

Beyond investment strategy, AI, encompassing ML techniques, plays a crucial role in bolstering risk management through the analysis of asset correlations and real-time portfolio monitoring. Furthermore, it contributes to the ongoing development of client interactions by enabling the delivery of more personalised financial advice tailored to individual needs and circumstances, which we wrote about last week!

The applications of these transformational technologies demonstrate that the initial wave of AI adoption across the wealth management sector has largely centred on providing advanced analytical support and automating routine tasks. The focus has rightly been on augmenting the capabilities of human professionals, providing them with enhanced tools to make more informed decisions rather than directly replacing their expertise.

As we progress along the ‘yellow brick road’ paved by data lakes, pipelines and large language models, we will see a shift from AI as a learning and informing tool, to AI as an autonomous executor.

…Or should I say task doer?!

Agnetic AI – From faciliatation to decision-making 

The emergence of agentic AI signifies a substantial step forward in the evolution of artificial intelligence. Unlike traditional machine learning, which primarily focuses on analysis and prediction, agentic AI possesses the capacity for autonomous action and independent decision-making.

These intelligent systems utilise a combination of machine learning, natural language processing (NLP), and reinforcement learning to not only understand complex situations but also to formulate and execute solutions with minimal human intervention. And therefore, Agentic AI systems are characterised by their ability to operate autonomously with limited human oversight, pursue defined goals, make proactive decisions, adapt to changing circumstances and seamlessly integrate with other tools and systems.

Sounds pretty amazing right, but think about the risk!

The integration of decision-making capabilities will allow AI agents to independently handle tasks such as processing financial transactions, reconciling discrepancies and optimising operational workflows in real-time. This represents a significant departure from the more passive role of traditional ML, which typically requires human professionals to interpret the results of analyses and then take appropriate actions.

Removing accountable and responsible people from decision-making in a risk-on environment? That’s crazy, I hear you say.

But maybe with the right training, testing and refinement, it might not be!

The development of intelligent AI “agents” capable of autonomous decision-making opens the door to the complete automation of intricate processes that were previously considered too complex for machines. These advanced systems can tackle multi-step problems through sophisticated reasoning and iterative planning, suggesting a transformative potential for the financial sector in the coming years.

Agentic AI, often considered the third wave of AI, builds upon the foundations laid by predictive and generative AI, with the ultimate aim of empowering human professionals by enabling AI to collaborate with them across various tasks and sectors at scale.

This next phase of AI is distinguished by its capacity for independent action and decision-making, marking a clear advancement beyond the analytical and predictive strengths of machine learning. The autonomous capabilities of agentic AI are underpinned by the synergistic integration of diverse AI techniques, including machine learning, natural language processing, and reinforcement learning, working in concert to achieve complex objectives.

It sounds like word soup, but what it means is better decisions on routine, rational tasks, more human time freed up for creativity and communications… and hopefully, a low cost to serve and better capacity to drive up AUM.

The winds of change are blowing, it’s either time to open your sails or be left behind.

A generational change for knowledge work – maybe?

Despite the growing interest in AI and its ability to deal with even more complex and nuanced tasks – and even though a significant proportion of organisations planning its implementation – a large percentage of executives still express a lack of trust in these autonomous systems.

Alongside the ambient nervousness in the space, practical challenges also exist, including a shortage of skilled professionals, the limitations of aging IT infrastructure alongside the heightened, and developing security risks associated with autonomous agents.

The fidelity, security and ethical implications, alongside the potential for bias within AI algorithms are particularly pertinent challenges. Underscoring the importance of robust governance and regulatory oversight.

Successfully navigating the adoption of ML, AI, genAI & agentic AI will depend on building and maintaining operational efficacy, client trust, adherence to regulatory requirements, mitigating risk and upholding stringent standards for data privacy and security.

That being said, we can imagine a future where agents have been trained, optimised and verified to work on diverse tasks such as:

• Adaptive identity verification
• Fraudulent activity indentification
• Tax implication guidance
• Adaptive savings & retirement planning
• Hedging strategies
• And much, much more

If you are thinking about taking your next steps with AI and need a consultancy to provide some external expertise into the heart of your organisation, get in touch, our consultants and technologists would love to chat through your challenges and offer up our expert view!