Let the AI Do the Experimenting

An AI agent framework called `autoresearch` can autonomously optimize tasks by operating in a continuous loop of experimenting, measuring impact, and iterating on code. This approach is applied to a marketing budget optimization problem, where the goal is to maximize revenue from various campaigns while adhering to a strict budget. The process begins by establishing a simple baseline solution and then deploying the `pi-autoresearch` tool to take over. The AI agent analyzes the initial code, defines the objective and constraints, and begins its iterative process to find a superior solution for campaign selection.