CityScope AI

The Problem Cities publish thousands of open datasets covering housing, transportation, environmental monitoring, public spending, and infrastructure. However, most of this data remains inaccessible to the public because analyzing raw datasets requires technical skills in data science and analytics. Citizens, journalists, and policymakers often want answers to simple questions like: Why are rents increasing in Zurich? Which neighborhoods have the most traffic accidents? Is air pollution getting worse? Yet answering these questions requires downloading datasets, cleaning them, analyzing trends, and building visualizations — tasks that can take hours or days. As a result, valuable public data often goes unused.

Our Idea We built CityScope AI, a generative AI copilot that lets anyone explore city data using natural language. Instead of navigating spreadsheets or dashboards, users can simply ask questions about their city.

CityScope AI then: Finds relevant datasets from open data portals Analyzes trends and relationships in the data Generates visualizations and charts Explains insights in clear natural language The goal is to transform raw public datasets into understandable insights in seconds.

Inspiration Our inspiration came from the growing availability of open government data and the realization that most people cannot easily interact with it. At the same time, advances in generative AI make it possible to build intelligent systems that reason over data and explain results.

We asked ourselves: What if anyone could simply ask questions about their city and instantly understand the data behind it? CityScope AI aims to become a universal AI interface for public datasets, helping citizens, journalists, and policymakers make more informed decisions.

How We Built It CityScope AI combines multiple components into an AI-powered data analysis pipeline.

Data Retrieval We use OpenData.ch datasets as the primary source for public data related to transportation, housing, and environmental metrics. To complement this data, Apify collects additional live information from web sources.

Data Processing Using Databricks, datasets are processed and prepared for analysis. This enables efficient handling of structured data and allows the system to identify trends and correlations.

AI Reasoning Layer A Hugging Face language model interprets user questions and determines: which datasets are relevant what analysis should be performed how results should be explained

Visualization & Interface Using Lovable, we built a simple interface where users can ask questions and receive: generated charts summarized insights AI explanations

The full pipeline transforms natural language questions → data analysis → visual insights.