Authority Index is the foundational data layer of 📝Authorix — a continuously scored graph of identity-resolved human experts ranked by how the expert community actually cites, follows, and engages with them rather than by follower count. Every Authorix surface reads from it.
It exists because semantic relevance systems rank text that sounds credible, and AI-generated text is optimized to score well in exactly those systems by design. The Index answers a different question — who actually knows this — by scoring the people behind content instead of the content itself.
It is consumed by researchers assembling ranked expert lists, journalists sourcing credible voices, marketers separating real practitioners from follower counts, and AI teams grounding retrieval in verified expertise.
This is 📝PersonRank built as a vendor product rather than inside a search engine, which is the part I find interesting — the authority layer is being assembled by a third party the engines can buy from. It carries the same discomfort I named there: it makes the web legible to machines by requiring everyone to show their papers.
Key Capabilities
- Cross-platform identity resolution — Social profiles, video channels, podcast appearances, papers, and editorial mentions collapse into one entity, requiring presence across at least two independent sources.
- Connectivity-weighted scoring — Authority compounds from high-authority connections on relevant topics while low-quality connections carry almost no weight, so raw follower count does not move the number.
- Bottom-up topic taxonomy — Tens of thousands of topics emerge as connection clusters densify, and existing topics merge back together when those connection patterns collapse.
- Signal growth rate — Each entity carries a velocity measure alongside its score, showing how fast its signal is moving relative to the average for that topic.
- Reported coverage — Authorix claims 3 billion-plus accounts analyzed, roughly 86% of global human authorities, 40,000-plus structured topics, across nine platform data sources.
Getting Started
- Browse the index free at authorix.com/explore, filtered by topic or country.
- Check a single person, handle, or URL at authorix.com/verify.
- Read how the scores are produced at authorix.com/methodology.
- Query it programmatically through the Authorix REST API.
- Connect it to an AI client through the Authorix MCP server.
Related
- 📝Exa.ai — a neural web-search index Authorix names as an alternative that returns snippets rather than ranked people.
- 📝Perplexity — named in the same comparison; answers with generated prose where Authorix returns indexed entities.