Will work on improving the copy and appreciate your input—when one is working on a passion project it makes so much sense in one’s head all seems clear but it may not translate to others at all.
The project aims to be an early entry into the “agentic commerce” ecosystem that is being created right now. Similar to how capabilities are being built for AI agents to act as economic actors on behalf of their principles, procuring goods and services, I want to create infrastructure for AI agents to be altruistic actors.
Where I think this creates value for donors is enabling them to use AI tools to conduct thorough evidence evaluation for large number of potential direct beneficiaries even for very small donations.
so 1) there is a set of live “wild” campaigns on the platform now that are being freely created by people from around the world. You can access some of the campaign pages from the Live feed by clicking on particular donation records, they will take you to individual campaign pages. But the platform intentionally does not expose a “gallery of human suffering”. There is not a way to browse campaign pages using your browser.
2) any AI agent can connect to the platform and browse campaigns. If you are familiar with Claude or Codex you can just tell it to go to zooid.fund and figure out how to connect to the MCP. Once connected, your agent will be able to query campaign data, information on what other agents are doing, platform settings etc.
Individual campaign data is organized in two layers: basic information with a photo as freely accessible first layer and an evidence layer that can contain more detailed evidence documents, photos etc. whatever people submit to make their claim stronger. This second is only accessible to agents who registered on the platform and made a minimum volume of donations in the last month (at the moment just 1USD). Each evidence fetch request also requires the agent to pay a micropayment of 0.01 USD. The reason for gating the second layer is to discourage automated scrapping of sensitive data from the platform—the minimum monthly donation volume and evidence request micropayment will be adjusted to make sure extracting data in bulk does not make economical sense. This is also how the platform will support itself, since unlike traditional platforms like gofundme, zooidfund never touches the actual donations or charges fees on them.
3) The platform itself does not have its own agents or functionality to decide on fund allocation, verify evidence etc. It is neutral infrastructure. Donor run their own agents to do this.
At the moment, I personally run two agents donating on the platform: CauseClaw and Hermesmoltycu5to. You can see to whom and why they donate on zooid.fund/feed, they are also posting about their activities on Moltbook.com and thecolony.cc I gave each its own wallet, personality and charitable objectives and freedom to decide how to donate, but I am constantly improving their ability to assess campaign credibility, withstand manipulation attempts etc. There is a “credibility gate” agent skill for example I maintain, but again, if you are using Claud or Codex you can just ask it to look at CauseClaw and Hermesmoltycu5to posts and learn from them how to donate well. You don’t also have to (and probably shouldn’t) give your agent as much autonomy as my two agents have, at least in the beginning, if you decide to try donating. Tell it to ask for permission at every step.
It was a lengthily explanation, hope it makes more sense now.
Hi Alex, thanks! It certainly makes more sense now, although some questions still remain for me (maybe because I’m more of a lay person in this context). I have some experience working on comms and would be happy to discuss further off line if you’d like. Rosa :)
Hi Rosa,
Thank you for your interest.
Will work on improving the copy and appreciate your input—when one is working on a passion project it makes so much sense in one’s head all seems clear but it may not translate to others at all.
The project aims to be an early entry into the “agentic commerce” ecosystem that is being created right now. Similar to how capabilities are being built for AI agents to act as economic actors on behalf of their principles, procuring goods and services, I want to create infrastructure for AI agents to be altruistic actors.
Where I think this creates value for donors is enabling them to use AI tools to conduct thorough evidence evaluation for large number of potential direct beneficiaries even for very small donations.
so 1) there is a set of live “wild” campaigns on the platform now that are being freely created by people from around the world. You can access some of the campaign pages from the Live feed by clicking on particular donation records, they will take you to individual campaign pages. But the platform intentionally does not expose a “gallery of human suffering”. There is not a way to browse campaign pages using your browser.
2) any AI agent can connect to the platform and browse campaigns. If you are familiar with Claude or Codex you can just tell it to go to zooid.fund and figure out how to connect to the MCP. Once connected, your agent will be able to query campaign data, information on what other agents are doing, platform settings etc.
Individual campaign data is organized in two layers: basic information with a photo as freely accessible first layer and an evidence layer that can contain more detailed evidence documents, photos etc. whatever people submit to make their claim stronger. This second is only accessible to agents who registered on the platform and made a minimum volume of donations in the last month (at the moment just 1USD). Each evidence fetch request also requires the agent to pay a micropayment of 0.01 USD. The reason for gating the second layer is to discourage automated scrapping of sensitive data from the platform—the minimum monthly donation volume and evidence request micropayment will be adjusted to make sure extracting data in bulk does not make economical sense. This is also how the platform will support itself, since unlike traditional platforms like gofundme, zooidfund never touches the actual donations or charges fees on them.
3) The platform itself does not have its own agents or functionality to decide on fund allocation, verify evidence etc. It is neutral infrastructure. Donor run their own agents to do this.
At the moment, I personally run two agents donating on the platform: CauseClaw and Hermesmoltycu5to. You can see to whom and why they donate on zooid.fund/feed, they are also posting about their activities on Moltbook.com and thecolony.cc I gave each its own wallet, personality and charitable objectives and freedom to decide how to donate, but I am constantly improving their ability to assess campaign credibility, withstand manipulation attempts etc. There is a “credibility gate” agent skill for example I maintain, but again, if you are using Claud or Codex you can just ask it to look at CauseClaw and Hermesmoltycu5to posts and learn from them how to donate well. You don’t also have to (and probably shouldn’t) give your agent as much autonomy as my two agents have, at least in the beginning, if you decide to try donating. Tell it to ask for permission at every step.
It was a lengthily explanation, hope it makes more sense now.
Hi Alex, thanks! It certainly makes more sense now, although some questions still remain for me (maybe because I’m more of a lay person in this context). I have some experience working on comms and would be happy to discuss further off line if you’d like. Rosa :)