A side project that turned into something other people might find useful.
If you use Stata, you know this routine: run your regressions, get results in the console, then spend a disproportionate amount of time turning that output into a LaTeX table that looks decent.
You can learn esttab or outreg2 (which have their own quirks), write LaTeX by hand (tedious), or use some GUI tool that never quite formats things the way you need.
It's especially annoying if you're a student learning econometrics and Stata at the same time, and now you also need to figure out LaTeX just to submit a formatted table.
tables.pub uses AI to read your Stata output and generate a properly structured LaTeX table from it. It identifies the command type, extracts the numbers, and formats everything according to academic conventions.
You can customize the output with dropdown options or just tell the AI what to change in plain English. The LaTeX is compiled on the server and you download the result as PNG or PDF.
No LaTeX installation on your end, no package conflicts, no configuration.
AI Parsing — A language model reads your Stata output and extracts coefficients, standard errors, significance levels, and statistics into a structured format.
LaTeX Generation — The structured data gets turned into clean LaTeX using booktabs and standard academic formatting conventions.
Compilation — A TeX Live installation on the server compiles the LaTeX into a high-res image or vector PDF.
AI Chat — You can describe changes in plain English and the AI modifies the table options or LaTeX code directly.
Built with Next.js, FastAPI, and Claude. The whole thing started as a weekend experiment with Claude Code.