Synthetic persona data is most useful when it reflects not just varied biographies, but the demographic relationships that shape them. This collection pairs six complementary descriptions with UK population and labour statistics, giving developers a way to generate or evaluate model interactions without representing real people.
What Sets It Apart
- Each of the 1 million records includes six persona narratives—general, professional, sports, arts, travel and culinary—alongside contextual attributes such as age, education, occupation and location. This makes it possible to select a population slice and examine how different interaction settings affect generated conversations.
- Its demographic grounding combines separate census-based probabilistic models for England and Wales, Scotland, and Northern Ireland. The records follow the adult population distribution and cover 372 local authorities, helping avoid the geographic and occupational narrowness common in generic persona sets.
- The dataset contains about 618 million tokens, including 242 million persona tokens, and is released under CC BY 4.0. It is designed for synthetic-data generation, model evaluation and user simulation rather than as a source of real user profiles.
Who Should Use It
A good fit for teams building UK-focused language models, testing AI agents across demographic and regional scenarios, or auditing synthetic-data diversity. Look elsewhere if you need minors, real identities, published ethnicity or religion fields, detailed addresses, or specialist finance and healthcare personas; those are not included. The records are artificial, and statistical alignment does not make them a substitute for evidence about real individuals.
