Skip to content

Chris Blattman Thread: From Claude Code Skeptic to Power User

Key Ideas

  • Blattman went from Claude Code skeptic to enthusiast in four weeks, despite not being a coder -- his use cases are project management, email triage, and team coordination
  • Built custom Claude Code skills: /morning-brief (email triage with draft replies), /checkin (urgent items + focus reminders), /weekly-review (transcripts + emails + chats into project dashboards and to-do lists)
  • The /write-proposal skill uses meeting transcripts, strategy discussions, and pre-analysis plans to draft proposals in his voice, reducing 12-hour tasks to 1-2 hours
  • Created claudeblattman.com to share tools, tutorials, and a GitHub repo for other researchers
  • Applications extend to field experiments, gang leader interviews, theoretical models of organized crime, and longitudinal panel management
  • Describes the experience as "transformative" and emphasizes accessibility for complete beginners

Summary

Chris Blattman, a development economist at the University of Chicago, recounts his rapid conversion from Claude Code skeptic to power user. His initial skepticism stemmed from not being a coder and seeing no relevant use cases. However, after experimenting with building custom tools for his actual workflow -- managing teams, triaging email, tracking projects -- he found dramatic productivity gains. His key innovation was creating Claude Code "skills" (custom slash commands) tailored to the recurring tasks of an academic researcher who manages multiple field projects.

The specific tools he built address common pain points in academic research management. The /morning-brief skill triages email and surfaces high-priority tasks with draft replies. The /weekly-review skill transforms meeting transcripts, emails, and chat logs into structured project dashboards and individual to-do lists, solving the problem of institutional memory in research teams. The /write-proposal skill leverages accumulated project context to draft grant proposals, reducing a 12-hour task to 1-2 hours. Blattman packaged these tools into a website (claudeblattman.com) with tutorials aimed at both beginners and power users.

What makes Blattman's account notable is the non-coding use case. Rather than using AI for data analysis or econometric coding, he applies it to the managerial and administrative dimensions of running a research program -- the tasks that consume substantial faculty time but are rarely discussed in productivity frameworks.

Relevance to Economics Research

Blattman's thread highlights an underappreciated dimension of AI productivity in economics: project management, team coordination, and administrative overhead. While most discussions of AI in economics focus on coding, data analysis, and writing, Blattman demonstrates that the managerial tasks surrounding research (email triage, meeting follow-up, proposal writing, team coordination) may offer equally large productivity gains. His approach also demonstrates that non-coders can build powerful custom AI tools, lowering the barrier to adoption across the profession.