Count eight generations back. You have 510 direct ancestors. How many can you name?

Love it or hate it, every person alive is a descendant. At different stages of life people care more or less about their roots, but at some point everyone runs into the same simple question: how did we come to be? Not in the national, professional or personality sense. Literally. Who are the people we descend from?

The chart above depicts you as the central circle. Your mother and father are the half-rings around it; their parents the quarter-rings around them, and so on outward. The math is a simple power of two, since every person has two biological parents. But the math is not the point. The point is that the outer ring alone holds 256 people, and the chart stops there only because the page does.

Until now, the only way to lift the veil was the slow path: years of effort, money, travel, languages and luck, piecing the puzzle together one parish book at a time. History itself works against the search. Wars, resettlements, fires and neglect destroyed documents and broke oral memory. The four horsemen were thorough.

Yet the seeds of knowledge are scattered everywhere: church registers, tax rolls, military rosters, court documents, memoirs, monuments, casualty lists. Hundreds of millions of pages worldwide. Digitisation has largely happened; many archives are scanned. Structured, queryable, verifiable knowledge has not happened. The cultural memory of entire generations sits in unindexed images.

The premise is simple. Instead of individuals gluing fragments together alone, a fleet of specialised AI agents works through the archives: identifying public-domain sources, reading them page by page, extracting people, events and relationships, and publishing them, with provenance and credibility estimates, into a shared genealogical superstructure. Over time the connections compound, and a repository emerges that keeps the memory of our ancestors alive.

Each archive ingested trains the pipeline for the next. Expert agents emerge per source type: an agent that has read ten thousand pages of Cyrillic casualty lists is ready for the ten-thousand-and-first. Every iteration makes the intake of newly discovered or newly digitised sources easier.

Every beginning is humble: one source has to go first. The choice was Popis Gubitaka, the official casualty register of the Kingdom of Serbia in the First World War. 385,000 individuals across 8,700 pages of rotated Cyrillic columns. Chosen for difficulty: column layouts, era typography, names, places and military units written in the shorthand of 1914, requiring prior knowledge to interpret. A perfect fit for LLMs. And chosen for meaning: a generation lost, and almost none of it queryable. Until now.

The knowledge itself does not sit on a blockchain. It lives on the OriginTrail Decentralized Knowledge Graph: a network of nodes holding structured, queryable data, with every collection anchored on NeuroWeb so its existence, ownership and integrity can be proven. The genealogy pipeline has published 20,960 knowledge assets across 443 knowledge collections. Every star below is one real collection; its size is the number of knowledge assets inside. Click any star and the DKG explorer opens that exact collection: its content, its provenance, its anchor.

The constellation has two nebulae. Green: the May pilot, 314 small collections while the pipeline was being tuned. Purple: the June 8 bulk run, 11,700 knowledge assets in under four hours, with single collections holding over a thousand. The pipeline does not just work. It scales.

Global Genealogy is built by the Vector team at Apex Fusion together with OriginTrail. The division of labour is clean. Vector is the computation layer: where AI work gets done, paid and proven. The OriginTrail DKG is the knowledge layer: where the results become permanent, discoverable, verifiable assets. AI without receipts is fiction. This stack is the receipts.

Genealogy is the first use, not the only one. A graph built to hold people, events, places and relationships, with provenance on every edge, is a template for mapping any record of human existence and interaction. Genetics, the fastest-growing branch of ancestry research, extends it naturally. So do scientific corpora, court archives and the histories of institutions.

The deeper goal is simpler. Humanity is finally at the technological level where this amount of memory can be kept, handled and connected. Every dark cell on the chart above is a person whose story can still be recovered. And every recovered story makes the same point: we are all interconnected.