Watching AI company staff on X and Reddit with public data only

A usage limit change can first become visible as a short public announcement, with an important qualification in a follow up reply. I want to find that information without turning a useful monitor into a collection of everything people say online. I am Jeremy Koski, an independent Computer and Information Research Scientist in Honolulu. My Python and PostgreSQL project, ai_resets, tracks AI usage resets, plan changes, promotions, and credits. Its X watchlist covers 113 handles associated with AI companies, including company accounts and staff. Reddit is another discovery source, with different identity and access constraints. The subject is public product information. The goal is to retain the announcement, its origin, and its qualifications, not to infer private activity or build personal profiles. This article describes how to define questions before the watchlist, keep identity evidence separate from announcement evidence, use an explicit import boundary for X, preserve thread context, give Reddit a different role, track publication time and collection health, and retain evidence with a clear purpose.