Paste any URL — website, YouTube, Instagram, TikTok. Each stage tries to extract a structured recipe. If one fails, the next kicks in. Proxy rotation, circuit breakers, and a failed-URL cache keep it alive at 99% success across 5K+ extractions.
01JSON-LDStructured Data
02Meta TagsOG / schema.org
03Direct ScrapeHTML Parser
04Browser RenderPlaywright
05Vision AIGPT-4V
06Text AIGPT-4
07FallbackManual Extract
Hover a stageEach stage handles a different content shape — the pipeline auto-routes to whichever extracts cleanly.