Been using it as the web-fetch layer for a coding agent setup instead of maintaining my own scraper/proxy stack. The markdown-cleanup output is genuinely good for feeding pages into an LLM without a bunch of nav/footer junk eating tokens, and the brand/company enrichment endpoints saved me from writing a bunch of one-off logo/favicon scraping code. Typed SDK meant it was a 10 minute integration like they claim, not a "10 minutes then 2 hours of debugging" situation.