Understand the city.
Base decisions on data and verifiable sources.
City Twin is an urban digital twin for analysing the city, testing interventions and building the case for investments, built on verifiable data and documents. The model of your city is built from public documents, open data and data provided by partner institutions: strategies, the PUG (General Urban Plan), the mobility plan, the budget, indicators. The assistant is configured to answer from the sources built into the city model, to point to the documents it used and to flag information that the data cannot support. Once the data is integrated, the platform produces a summary with the sources used, which a specialist checks before it is used.
How a question becomes an analysis on the map
The map uses the data of the demonstration model, version 0.3.0, updated on 2 August 2026. Below the video we explain what is measured, calculated or estimated. In the application the same feature is called “Analize pe departamente” (analyses by department): eight themes, each tailored to the questions of one department, with a legend and the data source alongside. The public demonstration opens two of them; the rest are for partner institutions.
From the model data: the historic centre (Centrul istoric) has 73.7% impervious surface, the highest share of the 15 neighbourhoods, and only 3% green surface; next come Florilor-Craiter with 60.5% and Astra with 59.4%. An estimated 24,326 people live within a 15-minute walk of the marked point, 11,462 within 10 minutes and 4,100 within 5 minutes. The values come from the demographic layer of the routing service, not from a count by address.
Each theme has a different source and method. Imperviousness is measured from satellite: Copernicus HRL Imperviousness Density 2021, 10 m resolution, average per neighbourhood. The other three are prioritisation indicators calculated in the platform, on a scale from 0 to 100: planting priority from mineral surfaces and mapped green spaces, charging station shortfall from built-up area compared with existing stations, solar potential from built-up area and annual irradiation. Neighbourhood boundaries, green spaces, stations and the street network: OpenStreetMap. The walking area is calculated along streets, not as a circle. The map is rendered with MapLibre.
A digital model of your community. Not a chatbot that knows everything.
An urban digital model brings the city's documents and data together in a representation that can be analysed and updated, that shows the coverage and limits of the data, and on which you can test decisions before taking them. City Twin works in a space dedicated to a single community, using documented data that comes with its sources. It is useful to the administration, to academia and to the community.
Configured for a single community
The assistant uses the community's data model: areas, projects, problems, indicators and sources. It is configured not to use information that is not supported by the sources built into the city model.
Answers from the community's data
Answers are based exclusively on the community's domain and data. Where information cannot be supported by the model, the assistant flags the missing data.
Control and protection for the institution's data
The data provided by the institution and the resulting model can be exported in open formats, under the terms set by contract and in line with the licences of the sources used. Primary hosting and AI processing take place in the European Union, and data protection is built in by design.
Sources first, then conclusions
Generative models can produce plausible answers even when the information is not sufficiently supported. City Twin is designed to tie answers to the sources built into the city model. That way an answer can be checked before it is used in a meeting, in a public document or in a funding application.
Answers point to their sources
The answer cites the document, layer or indicator it comes from. You can open the source to check it.
Flags when the data does not support an answer
Data gaps are flagged explicitly. Information that the data cannot support is flagged and must be checked before use.
Separates facts from assumptions
The platform separates measured values from estimates and shows how well each conclusion is supported. Estimates are not presented as measurements.
Verifiable arguments for a decision
It gives you options and arguments with sources; the decision stays with people. It does not replace the specialist, but gives them the data and arguments they need.
What the administration gains
The same public data, turned into time saved and decisions you can back up with documents.
Stronger funding applications
You build the case for European or national funding applications with vulnerable areas, indicators and cited sources. You argue with figures and documents, not just general statements.
Decisions based on sources, not opinions
Every answer comes with its source in the city's documents. You make the case for a project before the local council with verifiable evidence. The decision stays with people.
Test locations before deciding
You see on the 3D map the estimated spatial impact of an investment on accessibility (a car park, a school, a public transport stop) before allocating budget. Example: where to place a nursery so that as many children as possible can reach it within a 15-minute walk, calculated on the real street network.
From scattered documents to decisions backed by evidence
Four stages, automated or assisted, from collecting the data to analysing and presenting the results. Uploading documents, validation and interpretation need a person to step in.
The platform in use
Six recordings made directly in the platform; they are not reconstructions or mock-ups. The videos have no sound and do not start on their own.
Digital View, at three workstations
Three longer recordings made in the platform on the Brașov model; two of them continue on Cluj-Napoca. Digital View appears both as a presentation scene and as a working background. Traffic in the first video is estimated from the street category, not measured; the investment register in the second contains demonstration entries placed on a real city. The videos have no sound and do not start on their own.
The city's data remains the city's.
The administration needs to keep control over its data and the ability to export and reuse it. With City Twin, the data provided by the institution and the resulting model are exported in open formats, under the terms of the contract and in line with the licences of the sources. Third-party components integrated into the map (commercial traffic, satellite imagery, terrain tiles) remain under their provider's licence.
The data remains the city's
The model is exported in the open JSON format, with a public schema; the geographic layers in it use the standard GeoJSON format, so they can be reused in any compatible GIS application. The data provided by the institution and the resulting model can be exported under the terms set by contract and in line with the licences of the sources used, with no lock-in to proprietary formats.
Layers can be opened directly in QGIS
The city model is also published as a WFS service, at your city's address. The urban planner adds the City Twin layers in QGIS or any other GIS program like any standard source, with no manual exports and no files sent by e-mail.
Data can be exported and reused
The platform uses open standards and documented formats. Data and services remain interoperable, so you do not depend on a single supplier for critical infrastructure.
Data protection built in by design
Primary hosting and AI processing take place in the European Union. Some auxiliary technical services may operate globally, under the conditions described in the privacy policy. We do not use advertising cookies and we do not carry out profiling. We apply security measures benchmarked against the relevant NIS2 requirements, without certification or attestation.
Open the map now, without an account
Available without an accountThe public demonstration includes a selection of features: the city map, an analysis of heat vulnerability with its sources and declared limits, and a simulation of a cooling point. Free questions to the AI assistant, all themes and layers, scenarios, reports and exports are for partner institutions. The demonstration is not an official implementation by a city hall. The platform interface is currently in Romanian.
What partner cities get
available to partner institutionsPartner institutions get a dedicated City Twin space for the community, configured using local data, documents and priorities: AI-assisted analysis, saved and compared scenarios, collaboration between authorised users and report export.
- AI-assisted analysis on the city model: you ask in plain language and get an answer with sources, assumptions and data gaps flagged explicitly. It is not available in the public demonstration, including on the server.
- Simulator available in the institution's space: all 12 investment types, any number of investments, scenarios saved on the city's server.
- Report export: you generate a report for the meeting papers or for the funding application file.
- A dedicated space for the institution: authorised users, with roles and administration separate from the map.
The images give an indicative view of the main City Twin features. The values shown are examples; up-to-date data can be explored in the public demonstration.
Who uses it and what they get, concretely
A digital model that helps you analyse and decide
See problems and projects on the ground Partly available in the demonstration
Answers, problems and projects appear on a 3D map of the city: you can locate every piece of information. Easy to understand and to present in a meeting. The public demonstration shows the map and a single analysis; the city's problems and projects are for partner institutions.
Up-to-date indicators Available to partner institutions
Climate and air quality data is updated daily through an automated process, with the history kept and a monthly version archived. The update frequency is shown on each layer.
Investment simulator Partly available in the demonstration
You place investments on the map (12 types, from a nursery to a cooling point) and see which areas and which problems they cover. The public demonstration opens a single type: the cooling point.
Comparable scenarios Available to partner institutions
You save a simulation as a scenario and compare scenarios with each other. Scenarios are saved in the institution's dedicated space.
Automatic alerts Available to partner institutions
Based on the history of the data collected, the platform identifies trends and signals when an indicator is approaching a threshold (WHO guidelines, EU limits).
Built on your city's data
The platform is configured on the basis of your city's data, priorities and documents, not on a generic template. And the city's data remains the city's.
What city halls ask us
The questions that come up most often in our discussions with administrations, and the answers available on this page.
What data does City Twin use?
The city's public documents and data: the PUG (General Urban Plan), the mobility plan, strategies, the budget and indicators, plus the urban geometry mapped in OpenStreetMap. Climate and air quality series come in daily through an automated process, with the history kept and a monthly version archived. Coverage and accuracy grow as the city brings in more data.
Does the city's data remain the city's?
Yes. The model is exported in the open JSON format, with a public schema, and the geographic layers in it use the standard GeoJSON format, so they can be reused in any compatible GIS application. The data provided by the institution and the resulting model can be exported under the terms set by contract and in line with the licences of the sources used. No lock-in to proprietary formats.
Where is the data hosted and where does the AI processing take place?
Primary hosting and AI processing take place in the European Union. Some auxiliary technical services may operate globally, under the conditions described in the privacy policy. We do not use advertising cookies and we do not carry out profiling.
How do I check whether an answer is correct?
Every answer cites the document, layer or indicator it comes from, so you can open the source to check it. Data gaps are flagged explicitly, and what is measured is marked separately from what is estimated. City Twin is a decision support tool; the decision stays with people.
What can I try without an account?
The public demonstration includes a selection of features: the city map, an analysis of heat vulnerability with its sources and declared limits, and a simulation of a cooling point. Free questions to the AI assistant, all themes and layers, scenarios, reports and exports are for partner institutions.
What else do partner cities get?
A dedicated City Twin space for the community, configured using local data, documents and priorities: AI-assisted analysis, the simulator without the limits of the public demonstration, saved and compared scenarios, automatic alerts, report export and users with roles.
Which documents do we need to start a pilot, and what happens after the request?
The basic public documents are enough to start: PUG, strategies, budget. After you send the request, we contact you within three working days at most, we review the community's public data together, with no commitment, and then you receive a concrete proposal: what the digital model of your city would look like, the first report generated from its data and the terms of access to the full version.
Is the public demonstration a city hall project?
No. It is built from public data and documents, without the involvement of the city hall.
Would you like a digital model for your community?
The demonstration model was built entirely from public data and documents. We can start a pilot in the same way for any community that has the basic documents: PUG, strategies, budget. Tell us about your community and the objective of the pilot project using the form below: this is also where access to the full version begins.
What happens after you send it
Or write directly to [email protected].