True progress requires more than deploying isolated AI tools. Without a structured way to classify these systems, investment firms risk either over-relying on them or under-utilizing their potential, resulting in tactical improvements rather than transformative change.

Panthera together with others* have been working on a research initiative based on the central thesis that “design beats luck”: superior long-term performance stems from deliberately engineered decision environments, where human judgment and machine intelligence co-evolve to enhance decision quality (the ultimate key performance indicator in investing).

At the heart of our approach is a multi-dimensional AI taxonomy designed to bring clarity and governance to agentic systems. It classifies AI agents along three key dimensions: (1) their role in the investment process (idea generation, assessment, decision-making, execution, and monitoring, plus compliance); (2) the comparative advantage they strengthen (informational, analytical, or behavioral edges); and (3) the level of domain uncertainty.

Our taxonomy serves as both a governance tool and a strategic compass, helping investment organizations evolve into adaptive “decision ecosystems” that allocate capital more wisely and sustainably.

The CFA Institute published our article Design Beats Luck: How an AI Taxonomy Can Help Investment Firms Evolve (link)

*) The team consists of Ivana Zilic, Patrick J. Wierckx, Michiel Kühn, and Markus Schuller