Hey Product Hunt, I got tired of "best AI tools" lists where every tool is 4.8 stars and the top three paid to be there. So I built TrustRank. One number per tool: the share of members who would recommend it. That is the whole rating primitive. No stars, no five-axis composite, no editor picks. The one thing that moves a vote's weight is proof you use the tool. Upload a receipt or a dashboard screenshot, and your vote counts 3x. Not follower count, not account age, not who pays. Here is the uncomfortable part. Today there are 69 tools and 45 models listed, and not one of them has a score. A tool needs 3 reviews before TrustRank publishes a number at all, and 10 before it enters any ranking. So right now the leaderboard says "no reputation earned yet" and the reports page says "no issues yet." I could have seeded 50,000 reviews. Instead I wrote a test that fails the build if a number nobody produced reaches the UI. It caught a fake stat in my own homepage hero last week: "18,204 members voted", sitting there in the markup, wired to nothing. That copy is gone. An empty review site is a bad launch and an honest one. I would rather show you the empty version. What is enforced in code, not on a policy page: Founders cannot delete reviews. Claim your tool by publishing a token in DNS, a meta tag, or a file on your own domain. You get exactly one lever: a public reply with a founder badge. Criticism stays up. Ads cannot touch rankings. A guard test fails the build if the ads module is ever imported by scoring, ranking, search, the API, badges, embeds, or the public dataset. Reviews go through an AI check that can publish or escalate, never remove. Only a human removes a review, and every decision is logged with the thresholds it was made under. The data is downloadable. Full CSV at /data, with a tier column so you cannot mistake a withheld score for a zero. What I would love from you: Pick one AI tool you have actually used for more than a week and write the review you wish you had read before you paid for it. The unglamorous ones help most: n8n, Cline, Zed, Descript, Granola, Elicit. Three reviews turns a tool's score on. And if you think weighting votes by uploaded proof is naive, tell me why in the comments. I will be here all day. That mechanism is the thing I am least sure about. trustrank.so

TrustRank

AI tool reviews from people who proved they use them

AI tool reviews from people who proved they use them