Show HN: 10x token savings for AI Web Scraping

Category: ai-ml

Tags: web-scraping, memoization, browser-automation, llm-agents, cost-optimization

Score: 6.0/10 (Innovation: 7, Technical: 5, Documentation: 4, Utility: 6)

Makra proposes a memoization-driven approach to AI web scraping that caches page layouts to reduce token costs and hallucinations in agentic browser workflows. It claims 10x cost savings versus long-context LLM scrapers like Exa and Firecrawl, and ships a read-only browser harness for structured and tabular data extraction. The idea is compelling, but the project is an early-stage WIP with limited public code, benchmarks, and documentation.

Target audience: AI engineers and backend developers building agentic scraping pipelines

Repository: https://www.makralabs.org/about

View on Hacker News