These notes explore the principles behind AI-augmented investment research: where automation creates value, where human judgment remains essential, and how systems should be designed for trust, accountability, and long-term institutional advantage.
Most discussions of alpha focus on data, models, or capital. This note argues that the real constraint in fundamental investing is far more basic: time per idea. It examines why first-pass diligence has become the limiting factor in research throughput, and why addressing this bottleneck is a structural advantage rather than an operational optimization.
As AI systems become more capable, the temptation to automate judgment grows stronger. This note explains why, in investment research, removing humans from the loop does not eliminate risk but redistributes it in dangerous ways. It lays out the case for augmentation over autonomy and clarifies where accountability must remain human.
Speed without traceability creates fragility. This note explores why auditability is not a compliance add-on, but a foundational requirement for any AI system involved in research. It explains how provenance, logging, and reviewability are essential not only for regulators, but for institutional learning and trust.
Productivity gains are often celebrated without examining their downstream consequences. This note challenges the assumption that faster outputs are always better, and explains how opaque systems introduce hidden risk. It argues that productivity without inspectability defers problems rather than solving them.
General-purpose AI excels at breadth, but investment research demands precision, constraint, and accountability. This note explains why domain-specific systems outperform generic models in high-stakes environments, and how verticalization enables repeatability, trust, and long-term compounding of process knowledge.
Infrastructure choices shape governance. This note examines why ownership of AI systems — including where they run and who controls them — matters in institutional settings. It argues that client-owned deployments are not about distrust, but about aligning responsibility, authority, and accountability.
Financial returns compound, but so does organizational knowledge — if it is captured. This note explores why most research organizations systematically lose insight over time, and how systems that preserve context, reasoning, and evolution create durable competitive advantage. It reframes technology as a tool for remembering, not just accelerating.