Any pro cycling fan will know how fragmented the live race broadcast scene is. You have to check multiple websites, some dodgy, to find the race coverage times and sources. A real PITA. Course du Jour (coursedujour.com) does that for me and tells me daily "which bike races are on TV today, and where can I legally watch them." data collecting is the whole problem: ~30 sources, 5 languages, inconsistent naming, breaking weekly, sites with factual errors, etc. So AI does the upkeep: scrapers are dumb fetchers, and Claude reads the daily "unmatched race" report and opens PRs proposing fixes.
Another thing I hate doing: copying and pasting a new contact’s information into Google Contacts one field at a time: first name, last name, company, jost, email, phone, etc. So I built ContactFile (contactfile.app): privacy-first, no AI, all in-browser, no backend . You paste a messy email signature or LinkedIn blurb, get clean contact fields, one-click file to Google Contacts or download a .vcf. Pure regex + heuristics.