AI Adoption: Why South African Asset Managers Cannot Afford to Wait – Differentiation, not imitation The fundamental key to unlocking AI’s powerful potential within the asset management industry is to employ the technology as a differentiator, not as an imitator. We all have access to a range of AI tools—from machine learning, which learns from data to make predictions, to natural language processing, which understands and generates human language, to reinforcement learning, which improves through trial and error, and to generative AI, which creates human-like text, code, or even this article.
It’s essential to use these tools to create proprietary techniques that support the investment process—enhancing risk management, improving modeling and productivity, and most importantly, combining AI with human insight to augment expertise. The real edge still lies in our unique “secret sauce.”
Homogenisation risk
The next evolution of artificial intelligence—agentic AI, which has the capability to independently plan, select securities, and execute asset allocation without human input—introduces the risk of portfolio homogenisation across the industry. In addition to regulatory requirements mandating that investment firms must be able to justify and explain their decisions, this development could also lead to a trust-versus-fear dilemma, similar to the hesitation some passengers experience with the concept of pilotless aircraft.
Automating to accelerate alpha
At Matrix, we view AI not as a threat but as an augmentation tool. By automating repetitive, resource-intensive tasks, such as classifying and extracting insights from vast amounts of research, AI frees analysts and portfolio managers to focus on higher-order thinking—ultimately improving portfolio construction and risk-adjusted returns.
We have developed a proprietary generative AI tool, which enables staff to query internal research archives conversationally, as if messaging a colleague. This significantly shortens the time between question and insight, with built-in transparency and source traceability. The system is underpinned by retrieval-augmented generation (RAG), which combines generative AI with real-time data retrieval. Think of it as an open-book exam: instead of relying solely on pre-trained knowledge, the AI fetches relevant information from our document store before crafting a response.
This architecture mitigates common AI challenges such as hallucination by anchoring answers in verifiable content. While certain limitations remain (especially around temporal reasoning – AI’s ability to grasp and make decisions based on time-sensitive information), the overall gains in efficiency and coverage are undeniable.
What can AI be used for in the asset management space?
The above example of how we are using artificial intelligence to boost the productivity of the investment team, while managing for the risks that are still attendant with such technology, we know that AI can be used in many other aspects of the asset management industry.
It can, for example, assist in investment operations where greater efficiency and accuracy can lead to enhanced client experience as well as cost and time savings. One project in our pipeline involves automating the investor due diligence to support business development. Other possibilities include leveraging AI for compliance to manage regulatory complexity, reduce manual workload, and improve accuracy, or in risk management where AI can be used to monitor and manage risks across the spectrum of market to portfolio to operational risks. And, of course, manage portfolios, from selection to allocation to implementation.
The ethics of using AI
How ethical is it to use AI, and how much of its use should be disclosed to clients? What does it mean to be a responsible AI user?
The industry has long relied on models and early forms of automation—like rule-based systems, quantitative forecasting, and automated trade execution—that encoded human expertise into “if-then” logic. These tools helped support decisions but lacked the adaptive learning capabilities of today’s AI.
The real shift lies in AI’s ability to learn, evolve, and make increasingly complex decisions on its own. For example, unlike a traditional system that might sell a stock based on a fixed rule, a modern AI model could analyse real-time news sentiment, historical patterns, and market behaviour to decide when and how to act—without explicit programming for each scenario.
This raises new ethical questions around transparency, accountability, and trust. Clients may not need to know every technical detail, but they deserve clarity on how AI influences decisions that affect their portfolios.
We need to develop and incorporate oversight and compliance mechanisms that integrate legal, ethical, and operational safeguards to the use of AI in the work environment and we need to ensure full disclosure of its use. It is critical to ensure that trust is retained, even if there is no regulatory framework or if it is trailing the curve, because we cannot afford to lose client trust—you want to know if there is a pilot sitting in the pointy end of the plane if you will be flying in it. Ultimately the investment manager remains accountable for all decision making and outcomes.
A competitive imperative
While tools are emerging globally—such as those in beta testing at Bloomberg and Bank Credit Analyst—Matrix’s early adoption strategy gives us a tangible edge. Our ability to synthesise diverse data sources and deliver timely, actionable insights continues to strengthen our value to clients.
As generative AI becomes ubiquitous, the question for asset managers is no longer “if” but “how soon.” Those who delay embracing AI risk playing catch-up in a data-driven future that is already becoming the present.
AI’s evolution is inevitable, but how we engage with it will determine our success. In the words of one Matrix analyst, “Until we become machines, we must learn to collaborate with them.”
This article was originally written by a human and later revised using Artificial Intelligence (AI) to ensure clarity and technical rigour.
