Sphere
Independent prototype exploring agent-readable publishing, fragment intelligence, provenance-aware content, and the infrastructure of public knowledge in the age of AI agents.
About the project
Sphere is an independent research and prototyping project about agent-readable publishing.
As AI agents become readers, crawlers, researchers, and intermediaries, public knowledge needs to become more explicit: structured, licensed, provenance-aware, citable, and accountable.
Sphere explores this through the idea of a fragment: a bounded unit of public knowledge that can include content, sources, data, media, licensing, provenance, semantic relations, and access policy.
The first focus is Fragment Intelligence: helping publishers, authors, archives, and knowledge teams understand whether their content is ready for AI agents before thinking about live monetization.
What it does
- Transforms articles, documents, datasets, and media into agent-readable fragments.
- Preserves sources, provenance, licensing, and relations.
- Produces readiness signals for AI retrieval and citation.
- Simulates how agents might discover, inspect, and use fragments.
- Keeps publication separate from preparation through a private sandbox and publisher-controlled nodes.
Why it matters
The web was built for human readers, then crawled by machines. The next public sphere will be partly mediated by AI agents.
Sphere asks what a piece of public knowledge should look like when the first reader may be an AI system.
Status
Sphere is an early prototype.
The current work includes a public page, fragment format documentation, CLI tools, a validator, example fragments, an Agent Skill package, and a roadmap toward sandbox workflows, MCP tools, publisher intelligence, and payment-ready access.
Sphere is a personal, independent project.