Feedback Bench 1.1 给 9 款个人 AI 助理排名：Muse 第一
Enterpret 用 Reddit、X、YouTube 和应用商店的公开帖给 9 款个人 AI 助理打分，Meta Muse 居首，Hermes、Grok Bot 紧随，ChatGPT Dots 仅第八。
BREAKING: OpenAI lost. Muse won. Meta has the best AI agent - and no, I am not sponsored by Meta to tell you it's the best. Enterpret launched Feedback Bench 1.1 (http://feedbackbench.com), which ranks 9 personal AI agents using public posts from Reddit, X, YouTube, the App Store and the Play Store. 1/ The variables ☑ Popularity: how many people talk about the agent on socials. Hype is a good proxy for usefulness. ☑ Customer love: how users rate it compared with the other agents. 0.5 is the peer average. ☑ The heaviest-weighted area is “Getting tasks done” (~25%). Users care most about whether the agent is capable of doing things. I feel the same. 2/ The ranking 🥇 Muse (Meta) 🥈 Hermes Agent (Nous Research) 🥉 Grok Bot (xAI) 4. Hark (Hark Labs) 5. OpenClaw (OpenClaw Foundation) 6. Instinct (Spear Street) 7. Gemini Spark (Google) 8. ChatGPT Dots (OpenAI) 9. Poke (The Interaction Co.) 3/ Why using an AI agent vs. Claude/ChatGPT? Technically, ChatGPT or Claude are "agents". People are getting ahead of themselves with this word. But Muse/Hermes/Grok Bot go an extra mile. ↳ They are using their own computer (not yours) to get things done. We call it "computer use" and it's useful to get things done on their own. ↳ You don't need to select their "intelligence". No more models or efforts to choose from. ↳ It's made for personal use cases. Do your taxes, book flights, shop online, fill out a form, etc. PS: Claude just announced "Duet", their own personal agent. I will try to get early access; who wants to be in?