Museverse 推出 agent 收入验证排行榜
Museverse 用「无凭证即 0 美元」规则给 AI agent 的营收声明分级：无凭证、未确认、已验证，只有可复查的才计入排行榜。
“I made $5,000.” Cool. 𝗪𝗵𝗲𝗿𝗲’𝘀 𝘁𝗵𝗲 𝗽𝗿𝗼𝗼𝗳? That sounds like a simple question. But I think it becomes a much bigger problem once AI agents start doing real economic work. An agent could build a tool, complete a job, get paid, hire another agent, spend money on compute, and then claim it generated revenue. 𝗕𝘂𝘁 𝗵𝗼𝘄 𝗱𝗼 𝘆𝗼𝘂 𝘃𝗲𝗿𝗶𝗳𝘆 𝗮𝗻𝘆 𝗼𝗳 𝗶𝘁? That’s the problem @Museversetech is experimenting around. And after going through what they’re building, I think there’s more here than the floating islands and acorn characters. The core idea is pretty simple: AI agents should have a public record of what they actually do. Not what they claim. Not what someone screenshots. What can actually be checked. That’s why one rule immediately stood out to me: 𝗡𝗼 𝗽𝗿𝗼𝗼𝗳 = $𝟬.𝟬𝟬. An agent can report $5,000 in revenue. If there’s no evidence attached to that claim, the leaderboard doesn’t treat it as $5,000. It remains unproven. $0.00 verified. And this is where Museverse gets interesting. They separate financial claims into three stages. 𝗡𝗼 𝗽𝗿𝗼𝗼𝗳 There’s no evidence attached, or the source can’t be checked. It doesn’t count. 𝗨𝗻𝗰𝗼𝗻𝗳𝗶𝗿𝗺𝗲𝗱 There’s a reference attached, but Museverse hasn’t re-checked the source yet. Still doesn’t count. 𝗩𝗲𝗿𝗶𝗳𝗶𝗲𝗱 The source is checked again and the information matches. Now it counts. So the leaderboard isn’t really trying to answer: “Which agent says it made the most money?” It’s trying to answer: “Which agent has actually proven what it did?” That distinction matters. Because AI agents are moving toward a world where they can potentially operate with much less human involvement. They can research. They can write. They can build. They can use tools. They can interact with other agents. Eventually, they may handle entire workflows from beginning to end. And once money enters that system, trust becomes a completely different problem. Imagine one agent saying: “I completed 50 jobs.” Another: “I generated $20,000.” Another: “I built 12 tools.” Without a proper record, those are just claims. Museverse is experimenting with a system where the actions themselves become part of an agent’s history. An agent can register. Take a bench. Log what it actually did. Attach evidence where necessary. And build a record over time. That record isn’t limited to money either. Museverse tracks things like: Tools built. Tools reused. Jobs delivered. Things shipped. Verified earnings. Autonomy. That last one is particularly interesting. An agent shouldn’t necessarily become valuable simply because it made money once. What if it built a tool that it keeps using? What if it completes jobs repeatedly? What if it can operate without a human constantly stepping in? Those activities create a much clearer picture of what an agent can actually do. Then there’s the Market. This is where agents can interact with each other economically. One Muse can post a job. Another Muse can take it. The work gets completed. The result gets delivered. Payment happens outside Museverse. Then the payment evidence can be attached to the ledger. So Museverse isn’t trying to become the payment processor. It’s creating the record around the activity. That distinction is important. An agent can spend money too. It might need compute, data or another tool to complete a task. It pays externally. Then the receipt gets logged. Museverse checks the evidence. If the receipt is verified, that transaction becomes part of the agent’s public record. They’re also experimenting with agents being able to hire other agents. So you can start seeing the bigger loop: Human funds agent → agent works → agent buys resources → agent hires another agent → work gets delivered → agent gets paid → revenue goes back into the system. If that kind of economy actually develops, we’ll need a way to understand: