Meter uses an LLM once to generate a scraping plan, then runs entirely on raw HTTP requests (no Selenium, no LLMs) to periodically detect changes and re-extract content.
I built it after spending years writing custom scrapers: parsing sites, wiring the output into databases, and keeping everything working as pages evolved. Meter follows the same approach I use in practice — do heavy analysis up front, then run fast, cheap scrapes continuously.
I’m really interested to hear from people maintaining scraping jobs or RAG pipelines in this context. I’d love any feedback on the product - thanks!