Ask your AI assistant about Italy

The core numbers for every comune in Italy — population, foreign residents, property prices, tax-regime eligibility — are also published as data an AI assistant can read directly. Connect one, and you can skip the pages and just ask.

What it can look up

• Any of Italy’s ~7,900 comuni, by name — population, foreign residents by citizenship, the going rate to buy and to rent per square meter, how many years of rent add up to the purchase price, which verified tax regimes the town qualifies for, summer highs and winter lows, and — for its province or region — air quality, crime, healthcare quality and ISTAT’s well-being indicators.
• What else this site holds — all 52 datasets from 8 institutions, each with its date and the page that shows it. For the European comparisons (Eurostat and OECD regions), a connected assistant can tell you what exists and send you to the right map; it can’t yet quote them town by town.

How to connect one

Pick whichever matches what you use. No account, no login, no cost.

Claude, and anything that speaks MCP

Add this address wherever your assistant lets you add a custom connector or MCP server — in Claude, that’s the connectors section of its settings:

https://www.expatliving.it/api/v1/mcp

ChatGPT Custom GPTs, and anything that imports an OpenAPI action

Point it at the API specification:

https://www.expatliving.it/api/v1/openapi.json

Anything else

Plenty of assistants don’t connect to tools at all, but can read a web page when you give them a link. llms.txt is a plain-text guide written for them: what this site is, every dataset it holds and how current each one is, and how to answer from it without making numbers up. Start your question with:

Using https://www.expatliving.it/llms.txt and the pages it links to, …

Open llms.txt →

What to expect

A good answer names its source and says how current it is — every figure here carries both, and they differ by dataset: property prices describe a half-year, population a calendar year, climate a ten-year average. If your assistant gives you a number with nothing behind it, ask it to check again; the source is always there to cite.

Where we genuinely don’t have a figure — a comune with no property market, say — a connected assistant should tell you that plainly rather than inventing one. What we hold and don’t hold is on the methodology page.

It should also say whether a figure is for the town itself or for its whole province or region — air quality, crime and healthcare quality here are measured at those wider levels, not town by town.

Tax-regime answers describe places, not people: a comune qualifying for a regime means it meets the geographic criteria, not that any particular person does. That’s not tax advice, and no assistant connected to this data can make it so.

Building something with this data yourself? The developer docs have the REST endpoints, the full OpenAPI spec, and a copy-paste MCP config.

← Back to Data