How we built the British royal family

The trigger

The Habsburg demo had been live since May 2026. Beautiful, impressive, but a single dynasty. We needed a second proof — a family structured differently, in a different language, with a different history. The British royal family was the obvious candidate: internationally known, fully documented from 1066 to today, with its own complications that would stress-test our engine.

Day one — data structure

We started with William the Conqueror (born ~1028) and worked our way up to Charles III. First insight: the royal house isn't a line, it's a branching. The Tudors end with Elizabeth I, the Stuarts take over, then the Hanoverians via a distant cousin, then Saxe-Coburg-Gotha via Albert, and finally Windsor via a rename in World War I.

We built a seed script on Neo4j that creates the people and their parent-child relationships. 166 UUIDs, a family tree with clear roots and branches. First iteration: two hours of research, one hour of scripting, one hour of corrections. Then the graph went through.

Day two — life events and places

Phase two was place data and life events. Every person got: birth date and place, death date and place, and the most important stations in between. For monarchs that means: coronation, key battles, legislation, marriage(s).

We used Wikipedia as the primary source, with plausibility checks: every date cross-checked against the parent-child dates (no child can be born before its parents, no child later than nine months after the father's death), against their own lifespan (no coronation after death), and against genealogy consistency (whoever was king at one time cannot have been king somewhere else at the same time).

Our tree consistency checker found seven data errors in the first iteration. One date had slipped from the 11th to the 21st century. One coronation was five years after the burial. We corrected, the system went quiet again.

Day three — biographies and photos

Phase three was enrichment. For each person a short biography — three to six sentences, honest, no hero myths. Generated with Claude, then manually corrected, then stored. Clearly marked as AI-generated (mandatory per our writing style).

Photos: exclusively public-domain works from Wikimedia Commons — Renaissance portraits, Victorian photographs, official royal family portraits, all CC or PD. One image per person, with bounding-box info for the multi-person portraits (some Tudor family paintings depict six people in one frame).

What the engine does better with this family

Henry VIII was the first real test. Six wives, three surviving children from three different mothers. Mary I, Elizabeth I, and Edward VI are half-siblings. In most family-tree tools they'd be displayed as „siblings", losing two thirds of the story. Our engine shows: Mary and Edward are half-siblings on the father's side. Elizabeth and Edward are half-siblings on the father's side. Mary and Elizabeth are half-siblings on the father's side.

Victoria was the second test. Nine children, most married into European houses — Edward VII became British king, Alice of Hesse became grandmother to Tsarist and British heirs, Beatrice married into Spain. The engine computes that Victoria, via her grandchildren, was related to Wilhelm II of Germany, Nicholas II of Russia, and Alfonso XIII of Spain — all at once.

The third test was the Stuart-to-Hanover transition. Sophia of Hanover was named heir in 1701 because she was the closest Protestant descendant of James I. By strict Stuart genealogy there were closer relatives — but they were Catholic. The engine cleanly shows that Sophia was a fourth cousin to the crown, and that the line branches exactly there.

What we learned

Three days for 166 people means: just under 90 seconds of work per person. That's possible because our tools fit. Without AI bio generation it would have been more like ten minutes per person. Without the consistency checker we would have discovered errors only in the live demo.

What the demo has shown us since: people click through. They follow lines. They find relationships they didn't know existed — that Wilhelm II of Germany was Victoria's grandson, that Charles III is a direct descendant of William the Conqueror, that Diana Spencer was related to the Tudors via a side line.

That's exactly what Kinverse is meant to do: make visible the connections that get lost in compartmentalised thinking.