AI Agents Will Upend Investing Even More Than Coding - Frank Long, Coatue

Frank Long from Coatue outlines a framework for deploying AI agents in investment, conceptualizing a hedge fund as a function that transforms external and internal data into specific portfolio decisions regarding selection, size, and timing. Success depends on selecting the optimal "runner"—whether deterministic code, statistical models, LLMs, or human judgment—for each sub-task to maximize quality. This approach requires rigorous management of a vast, complex search space of hyperparameters, ranging from data collection and filtration to analysis and execution. Ultimately, building an effective AI-driven investment system involves creating a state machine capable of processing disparate information sources to make coherent, high-conviction bets, a challenge that extends beyond finance into fields like drug discovery, policy, and corporate strategy.