Show HN: 10x token savings for AI Web Scraping

I have been working on this problem for 6 months. Makra brings down the cost of web scraping by quite a lot, using novel memory and layout caching mechanisms. Currently it's web scraping, but the same can be applied with all kinds of Browser-use application. Let me know what you think. And, just FYI: benchmarks will be public soon. We're on it.