Blog / 21 Sept 2026
Mosaic Theory at Scale
A working paper on how AI can investigate cross-company relationships across thousands of companies without producing plausible stories at scale.
Most uses of AI in equity research start from the same question: how can an analyst do their existing work faster? This paper asks a different question: what becomes possible when AI has enough reading capacity, memory and analytical attention to investigate relationships across thousands of companies?
Mosaic Theory at Scale sets out the research system needed to do that without simply producing plausible stories. It covers reliable retrieval, explicit economic mechanisms, versioned research skills, relational memory and matched evaluation, then works through three examples to show what the evidence can tell an investor and where its limitations remain.