Presenter: Cory Baird
Affiliation: University of Tokyo.
VIMM Lab Seminar
Date: November 24, 2026.
Time: 15:00 Israel Time.
Paper, Slides and/or Data: TBD
Abstract: LLMs can generate code and economic analysis at scale, but verifying and reproducing the output remains a major challenge for empirical and theoretical macroeconomics. Part 1 covers current AI-related jargon alongside a brief overview of historical developments. Part 2 examines infrastructure design patterns (via Object-Oriented Programming and Cookiecutter design) and open-source tooling (such as open-source theoretical and empirical macroeconomic models), which are necessary conditions to keep research verifiable while retaining workflow efficiency and scalability. Part 3 covers practical workflow optimization using agent harnesses (Pi, Pydantic, Claude Code, Antigravity) alongside multi-agent orchestration with Herdr. The lecture concludes with a live coding demonstration and open-source libraries for benchmarking agent workflows in macroeconomic applications.