Using agents to build economic datasets

Short summary

DRIL (Deep Research on a Loop) is a methodology using AI agents to automatically assemble economic datasets from publicly available sources, reducing the cost of dataset construction in empirical research. The two-stage architecture applies a fixed research instrument across mapped unit spaces like countries-by-years. This approach scales to any structured data-collection task requiring systematic application across multiple dimensions.

• •AI agents can automate expensive dataset construction from public sources
• •DRIL uses two-stage architecture with fixed research instruments across mapped unit spaces
• •Applicable to empirical economics, research automation, and agent-driven workflows

Generated with AI, which can make mistakes.