In 2020, during peak COVID, I taught an introductory class for early-career social scientists over Zoom. I found that, although these researchers knew what experiments to run and how to design surveys and experiments, their coding skills weren't at the right level. They would spend a lot of time just installing libraries and debugging code. Now that we have GenAI, I think these researchers should focus on the research design and interpreting it, deciding what to run and so on, and not on code. I built Inquiro to fix this. It's a notebook workspace where you describe a research question, and it generates Python code with the right statistical method, diagnostics included. The key design choice: statistical methods are encoded as "skills" (difference-in-differences, regression diagnostics, RCTs, etc.), each one capturing what a good methods professor would say. The agent loads the relevant skill before generating anything.

Inquiro

Notebooks that turn data into publication-ready analysis

Notebooks that turn data into publication-ready analysis