webHarvest lets you describe what data you want off any website in plain English, no manual scraping scripts. It builds the extraction, tests it against the live page before you save anything, then runs on a schedule and tells you when something actually changes. AI planner reads the page and writes real, verified selectors (tested against every matching item, not guessed from one example), a real headless browser executes it.
Hey Product Hunt 👋
I built webHarvest because I kept manually checking a competitor's pricing page and wishing something would just tell me when it changed. Every scraping tool I found either needed real coding, or broke the moment a site's layout shifted slightly.
A few things I'm proud of:
You just describe what you want ("get the price and address for every listing"), the AI generates the workflow and tests it live before you ever save it
It tells you when extraction breaks instead of silently returning garbage data and if that happens, that run doesn't count against your usage
Real change detection it diffs each run against the last one and flags what's new, removed, or different
If a site shows a CAPTCHA, you get a live view of the actual browser and can solve it yourself, then hit resume
Would love your feedback and what would you point this at?