12 — AI-Assisted Research
Using LLMs effectively in research workflows.
This module covers AI-assisted research workflows: giving LLMs project context via CLAUDE.md and AGENTS.md, building reusable skills and slash commands, autonomous code review, and research life hacks including Obsidian and Git for everything.
Lecture slides
Code
Check the course repository for the R scripts.
The hands-on exercise merges Colombian customs import data with department-level 2018 election results to produce five figures and a summary table. Results are written up in Overleaf.
Key packages: data.table, ggplot2, sf, rnaturalearth, patchwork
Further resources
- Chris Blattman’s claudeblattman.com — AI workflows for researchers and managers
- Blattman’s GitHub skills repo — open-source skills and templates
- Claude Code documentation
- Obsidian — markdown-based knowledge management