Institutional investors remain cautious about integrating AI technology into the workplace. They’re exploring the possibilities step by step, looking to add a digital “sparring partner” to their investment teams.
Every technological revolution comes with a period of watchful waiting, said Christiaan Tromp of Ipfos. “The hesitancy we’re seeing from many Dutch institutional players around adopting AI in the workplace is no different from when the PC or the internet were introduced,” he explained. “But this is the perfect opportunity to onboard an increasingly intelligent ‘super assistant’.”
Opinions differ on what role AI technology should play in the investor’s workflow. Asset manager Schroders is all in on AI and has rolled out its own chatbot, Genie, which operates under strict security protocols. For private investments, the firm uses a tool called Gaiia that accelerates the creation of investment memos. “We’ve integrated AI into our business processes and are actively training our employees to use it as effectively as possible,” said CIO Nils Rode. “The system acts as an assistant to our analysts, especially when it comes to the heavy lifting in data analysis.”
Asset manager Comgest takes a more cautious approach. “We first want to better understand the risks associated with using generative AI in particular,” said Lodewijk van der Kroft. “Our impression is that it’s especially useful for firms with a quantitative or scientific approach to investing. We see investing as a balance between science and human creativity—and we definitely lean more toward the creative side.”
GenAI
Generative AI, or GenAI, is a form of artificial intelligence that not only interprets information but also creates new content autonomously. Schroders’ Data Insights team has been experimenting with it for the past three years. “We’ve seen significant efficiency gains, especially when it comes to qualitative information,” Rode explained. “Thanks to external large language models, we can now draft preliminary investment memos for hundreds of private investment opportunities at lightning speed, which saves time and allows for deeper analysis of more potential deals. The main strength of generative AI lies in qualitative data (text). For quantitative data, the system still tends to be more error-prone, but we expect that technology to improve quickly.”
Comgest also uses generative AI tools such as Claude and ChatGPT but, like Schroders, imposes clear boundaries. “We absolutely want to prevent any sensitive information from leaking,” Van der Kroft emphasized. “That’s why we work with an internal tool and conduct research in a secure AI environment to assess whether information remains protected and how AI can safely be used in that context.”
Tromp sees the generative power of AI as a real opportunity for institutional investors. “You can fine-tune your portfolio more precisely by asking the machine how certain objectives can be achieved. If it tells you, for instance, that you should invest one hundred percent in gold, that at least sparks a conversation and invites critical evaluation.”
‘Snap capacity’
The quality of the AI machine’s output still depends on the quality of the input provided by humans. Tromp refers to this as the investor’s “snap capacity”—their ability to formulate and process information clearly. “First, you need to clearly define your goal, then give the AI precise instructions, and finally, carefully evaluate the output,” Tromp explained.
“If the input is flawed, AI can replicate and amplify those errors. The key is to integrate AI carefully into the process without completely handing over the reins,” said Tromp. “It’s a continuous process in which humans must keep challenging the AI model. After all, the machine will never be able to predict the future.”
Rode noted there’s often a misconception that AI will replace humans. He sees the technology instead as a lever. “Human capabilities are enhanced, but humans always remain in control. It is important, however, to stay aware of the machine’s limitations. That’s something we actively train our people on.”
More dynamic and smarter
With AI, investors gain a valuable sparring partner that helps visualize scenarios, risks, and portfolio behavior. Tromp highlights that scenario generation and simulations using Explainable AI (XAI) offer direct support and transparency. “You can ask the machine unlimited questions, and it will explain how it arrived at a particular answer. This allows for endless sparring, without the bias of human opinions or the group dynamics of an investment committee.”
Tromp points out that demand for transparent and integrated XAI solutions is rising rapidly. “These solutions are essential for the responsible and scalable use of AI in asset management. They help firms make better decisions and explain those decisions clearly to clients—something that’s crucial for trust and regulatory compliance.”
Currently, asset allocation studies—also known as ALM studies—are often conducted only once every three years. These studies work within bandwidths for each asset class. “You can never be sure whether your asset allocation is truly optimal at that moment. With AI, you can reassess more frequently and with greater flexibility, making your portfolio more resilient instead of merely robust.”