Navigating the Challenges of a Rapidly Evolving Landscape

Download PDF

navigating-the-challenges-april-2026

The AI landscape is evolving at a pace unlike anything our industry has seen before. Today’s leading models are quickly surpassed, and new capabilities become standard almost overnight. In systematic investing, this evolution includes the rise of agentic AI – systems that can go beyond assisting with isolated tasks and can instead execute parts of the research process end-to-end. This includes exploring ideas, writing strategy code, and evaluating results, all while operating within human-defined constraints.

The challenge is not access to AI, but how to build something durable on top of it. It can be tempting to develop workarounds for the shortcomings of today’s models, but those gaps tend to close fast. What is custom-built today is often delivered as a built-in feature tomorrow.

To address this, our focus is not on short-term fixes but on integrating AI into the parts of our process that matter most – our data pipelines, research workflows, and back-testing systems. These are areas where domain expertise creates lasting value, and where external providers cannot offer off-the-shelf solutions. By building integrations that work seamlessly across models, we can create a framework that grows more valuable over time – improving with each new generation of AI.

However, the ability to switch models quickly on the technical side is only part of the equation. Without aligned legal and compliance processes, adoption can quickly become a bottleneck. Effective use of AI therefore requires coordination across the organization. Our legal, compliance, and technology teams work together to pre-approve leading providers, ensuring that new models can be deployed without delay. This allows us to fully benefit from the pace of AI development while maintaining strict security standards.

In the context of systematic investing, speed also introduces a subtler risk. When strategies can be generated and tested far more rapidly using AI, it becomes easier to uncover patterns that appear compelling but are in fact statistical coincidences. This makes human expertise essential to distinguish genuine signals from noise, and to apply informed judgment, specialist knowledge, and robust validation techniques.

We have seen this dynamic before. When machine learning first emerged in finance, there was a widespread belief that simply feeding data into a model and letting it find the answers would be sufficient, and that deep domain expertise was no longer necessary. That proved to be wrong. Today, similar ideas are resurfacing with the rise of agentic AI, suggesting that the technology is so powerful that it can replace human judgment altogether. We respectfully disagree. The ability to generate ideas at scale is transformative, but without experience to guide what questions to ask, how to validate results, and when to remain skeptical of outcomes that appear too good, speed simply becomes a faster way to find things that are not really there. In this context, our 25+ years of experience becomes more important than ever.

Working with AI means trusting external technology with sensitive information, and we treat that responsibility with the seriousness it deserves. Every provider and every point where data interacts with a third-party system has been carefully evaluated to ensure that data remains within its intended boundaries. Each AI agent operates in its own contained environment, with access strictly limited to what is required to perform its specific task.

The result is an approach that fully embraces what AI makes possible, while remaining grounded in the discipline and rigor that systematic investing demands.