Launching today
Feedback Bench by Enterpret
AI agents, ranked weekly by what users actually say
1 follower
AI agents, ranked weekly by what users actually say
1 follower
Feedback Bench ranks AI agents by what users actually say about them. Our first bench covers coding agents. We read 287,903 posts from 101,338 users across Reddit, X and beyond, and scored 17 agents on 63 criteria in 9 areas. Benchmarks tell you what's possible. Users tell you what's real. Compare any two head to head, see where each wins on price, limits and reliability, and find the best fit for how you work. New rankings every week, with personal agents up next. Built by Enterpret.
Hey Product Hunt 👋 my name is Jeff and I'm with the team here at Enterpret.
Everyone has a favorite coding agent, and everyone is pretty sure theirs is the best. Benchmarks don't settle the argument. They tell you what an agent can do on a test set, not what it's like to live with one every day, with real limits, real sessions and a real bill.
Making sense of what customers say is what we do at Enterpret, so we pointed that at coding agents. We read 285,697 posts from 98,493 developers about 17 agents, sorted everything they said into 63 criteria across 9 areas, and scored every agent on each one.
A few things surprised us. Price tracks how much people love an agent more closely than anything else we measured. Some things almost never earn praise no matter how well they work, like getting billing right or not getting stuck in a loop. And the agent at the top changes depending on what you care about, which is why you can compare any two head to head and see exactly where each one wins.
Rankings update every week as new posts come in.
Coding agents are just the first bench. Personal agents are next, and we plan to cover more AI categories after that.
Tell us which agent you use and whether the rankings match your experience. If something looks off, we want to hear it.
- Jeff