The Scraper Idea Is Boring. The Chargeback One Is Actual Money.

12 new cards today, 31 fresh signals out of a 9,870-signal web_search firehose. Most of it is the usual "AI wrapper for a thing" noise. Three of them made me actually stop and read the whole card.

Let's start with the one everyone's going to build, because it scored highest: Scrape any website without writing a line of code, 75 points, straight off Product Hunt.

And honestly? The pain is real. I've lost entire afternoons to "I just want this one table" — fighting selectors, pagination, Cloudflare, and a scraper that breaks the moment the site ships a redesign. But the market is packed. Browse AI, Apify, Octoparse, Firecrawl, a dozen Chrome extensions. A 75 score here is a score for the problem, not the opening. Freemium at ~$19/mo tiered by volume means you're competing with products that already have years of anti-bot infrastructure and a free tier that's genuinely good. If I built this I'd niche the fuck down — "scrape any real estate listing site into a spreadsheet" and own that one vertical. Generic no-code scraper is a VC-shaped hole, not an indie one. Hard pass, unless you've got a very specific vertical and a very specific buyer.

The one that actually made me sit up: Automated Chargeback Evidence Generator for Stripe, 71 points.

This is the good kind of boring. Nobody wants to build it, nobody wants to think about it, and every merchant who gets hit with one has to dig through order data, shipping records, and support chat within a 7–21 day window and hand-write a rebuttal. That's a fire drill with real money attached. And the monetization is chef's kiss — $19–49/mo tiered by dispute volume, or 15–25% of the recovered amount. Percentage-of-recovered is the play. It re-prices you against the merchant's actual loss instead of against other SaaS tools. First dispute free as the hook is smart too: they only feel the pain after they've seen a win.

Would I build it? Yeah, probably. The Stripe API does most of the work — pull charges, pull dispute objects, pull order + tracking + comms, assemble a PDF with the evidence fields Visa/Mastercard actually want. Ugly, unsexy, sticky. The risk is that Stripe eventually just... does this. They've been creeping into more of the merchant ops stack every year. But Stripe has been "about to kill that" for five years and small merchants still lose disputes to lazy evidence. I'd build it in a weekend with Next.js, the Stripe API, and a PDF library — the moat is knowing which evidence fields win which reason codes, and that's just domain knowledge you accumulate.

Then there's One-screen MRR and churn dashboard for Stripe-powered SaaS, 68 points, $19–49/mo.

Look. This is the third idea on this list and it's the same category twice. MRR dashboards are the "hello world" of indie SaaS. Baremetrics, ChartMogul, ProfitWell (free!), and about 400 abandoned side projects on GitHub. The card's own pitch — "MRR and churn data is scattered across Stripe, Google Analytics and spreadsheets" — describes a problem that Baremetrics solved in 2013 and now gives away. A 68 score is the algorithm being polite. If you're going to touch this space you need a genuinely different angle (true profitability across all costs, not just Stripe revenue, which is the other card at 66 — closer to interesting but still crowded). Don't build the dashboard. Build the thing that tells you which customers are actually unprofitable, including infra and support time. That's the version I'd pay for.

If I had to pick one thing off today's board to actually ship this week: chargebacks. Boring, painful, percentage-of-recovery pricing. That's the one where the customer's alternative isn't a competitor — it's losing $400 and an afternoon.