Every person in Europe, with a household, a neighbourhood and a wallet
Humanity is a synthetic population of Europe at one-to-one scale. Not a sample, not a set of averages: one record for every resident, placed in a specific square kilometre, inside a specific household, with a budget that adds up. Every one of them is generated from what the statistical offices publish, so the totals are right, the combinations hold, and nobody in it is a real person.
- 522 Mpeople, one record each
- 43country builds, the UK as four nations
- 304regions, every municipality inside them
- 1 km²the grain every household is placed at
- 700+attributes per person, household and place
- 47COICOP budget lines per household, and they always add up
Where you live says a lot about what you spend
Take two households with the same income. One lives in a 1930s flat, two minutes from a tram, with a café on the corner and nowhere to charge a car. The other lives in a house from 2004 at the end of a bus route, with a driveway and a fossil boiler. They earn the same and they spend on completely different things, and every model that starts from a national average gets both of them wrong.
Humanity starts from the place. The square kilometre is measured first, then households are formed inside it until the census counts are exact, then each person is drawn to fit the household and the neighbourhood. Income sets the budget, the budget sets the spend, the spend sets the emissions. Nothing is averaged across a layer. Nothing can contradict the layer beneath it.
So the three layers are joined, not merely stacked, and any attribute can condition any question in any other. Renters in their thirties with children, within walking distance of a tram, who spend above their neighbourhood on eating out is one query, answered with a count, a map and a profile. Not a modelling project.
What one row holds
The place
about 130 attributes per square kilometreNot a postcode lookup. Every cell is described from its own geometry, so the description exists at the same grain everywhere, including places no commercial dataset covers.
- How far to the city centre, the nearest station, grocery shop, school, hospital, motorway exit and fast charger, measured around water and across borders
- What is within walking distance: restaurants, cafés, bars, shops, transit stops, and how diverse the cuisine is
- Land use, building age, night lights, green space, the share of homes that can charge a car at home
- Heating and cooling degree days, sunshine, snow, hot days
- Who the neighbours are, from the 2021 census grid: age bands, employment, where people were born, how much the area churns
- How the zone fills up or empties by day
The household
about 400 attributes, plus a 47-line COICOP budget and a 55-line emissions ledgerThe unit most decisions are made in. This is the layer that answers whether they can afford it, and what buying it would displace.
- Gross, tax, net and equivalised income. Transfers and benefits. Capital income
- Wealth and debt as separate things: property, deposits, financial assets, mortgage, car and consumer loans, the rates on them and what servicing them costs
- Tenure, housing cost, energy cost, vehicle cost, and how heavy the housing burden is
- A complete spend budget across the 47 COICOP categories that always sums, with the online share and promotion sensitivity of each
- Resilience: months of reserve, what a rate shock or a job loss would do, whether a car or a boiler could be replaced without something else breaking
- The installed base: cars and their powertrain, heating system, solar, a second home. Emissions attributed to all of it
The person
about 160 attributes per individualWho they are and what will reach them. Every member of a household is a full row, linked to the same home and the same square kilometre.
- Age, sex, civil status, education, work, occupation, personal income and wealth
- Origin and cultural identity, drawn from 323 curated profiles, with a calendar of the events that move demand
- Around sixty attitudes from the European Social Survey: trust, values, satisfaction, politics, outlook
- Media diet, digital engagement, and paid subscriptions modelled together, so video, music, gaming, news and fitness co-occur the way they do in life
- Brand affinity: premium, sustainability, novelty, wellness, experiences
- Commute, weekly rhythm, daytime availability. The chance they move, change job, partner or have a child in the next three years
752 fields that hang together
This is Lukas Brandstätter, one record from the Burgenland unit. The name and the photo are labels we gave the row; everything else is the record. Notice how the place explains the household and the household explains the person: 33, professionals, rents at market rate, cycles 6 km to work because the station is 249 m away, spends 61% of net income and puts the rest aside, pays for video and music but not for news.
Open the whole row
- Household net income
- €83,100
- Yearly spend
- €49,800
- Net wealth
- €26,900
- Months of reserve
- 10
- Online share of spend
- 50%
- Nearest rail
- 249 m
- Heating degree days
- 3,054
- Household footprint
- 23.4 t CO₂e
- Chance of moving, 3 years
- 37%
- Paid subscriptions
- 2
Built to be leaned on
Anchored, checked, and honest about the edges. Every number here can be traced to the statistic it came from, and the ones that can't are marked rather than invented.
- Anchored at one kilometre Populations start from the measured census grid, and households are formed until the person counts are exact. Add up any region and you get the figure the statistical office published.
- Conditioned, not correlated Income before budget, budget before category spend, spend before emissions, along an explicit dependency graph. The layers cannot contradict each other.
- Provenance on every attribute Each one says whether it was measured and raked or modelled as a propensity. Where a country has no source, the field is left empty and marked. Nothing is borrowed from a neighbour.
- Checked against data it never saw Twenty-one automated gates run on every unit and their results are kept, including the failures. Held-out comparisons are published either way.
- Synthetic by design No row corresponds to a real person, so there is nothing to re-identify and no privacy obligation to carry. Results can be used, joined and published.
- Money set region by region Income, spending and emissions follow the region a household lives in, so Bucharest earns nearly twice the Romanian average and its people know it.
The long version, with the numbers: Accuracy, Sources and What's inside.
43 country builds, one method
The EU, the EEA, the UK as England, Wales, Scotland and Northern Ireland, Switzerland, the Western Balkans, Iceland, and the far corners: Greenland, Svalbard, the Azores and Madeira. Every country is built the same way on the same schema, which is what makes Lisbon and Luleå comparable, something the statistical offices themselves do not offer.
Not yet served: Ukraine, which has no household composition source and is declared unbuildable rather than estimated; Bosnia and Herzegovina, Moldova and Kosovo. The full list, with flags.
v2.1 today, and it keeps getting better
Humanity is a living model, not a dataset that was shipped once. Every Hiasynth answer today comes from v2.1, the first release built for customers. The next build, v2.1.2, is already under way: it rebuilds every unit on one consistent pipeline and adds a bio layer, with a birth date for every person, mother's age at birth and birth-month seasonality drawn from Eurostat. Every release is versioned and documented, so you always know which build answered you.
The lineage of HumanityAsk it something
Humanity answers from inside the AI assistant you already use. Connecting takes a few minutes.