We partnered with the City of Prague to develop Urbanora, an AI-driven solution designed to streamline and automate the reporting of city-wide issues. This initiative is part of the EU-funded program, “Communicity”, where we worked closely with Prague’s OICT organization to ensure the solution aligns with the needs of Prague's citizens.

• Enhance communication channels between Prague’s citizens and the municipal administration.
• Streamline and automate the reporting and resolution processes for city-wide issues.
• Ensure data privacy and compliance with GDPR and municipal guidelines by anonymizing sensitive user inputs.

Změňte, Prague’s existing citizen-engagement platform, allowed residents to report city-wide issues, which were then manually reviewed and routed to relevant city departments by a dedicated administrative team. This manual approach introduced delays and operational challenges.

To address this, we built Urbanora, an open-source, fully configurable AI-driven solution integrated directly on top of Změňte, automating the entire issue-reporting workflow with advanced AI features.

Our Solution included an automated classification system that categorizes citizen-reported issues, allowing administrators to route them to the appropriate city authorities’ quickly. We used multimodal large language models (LLMs) that automatically generate descriptions matching the uploaded images and identify issue categories, significantly streamlining administrative reviews.

To ensure inclusivity, the platform supports multilingual submissions by translating citizen reports from various languages into Czech. Recognizing the sensitivity of user-submitted data, the solution incorporates anonymization techniques to blur sensitive details, such as license plates, personal addresses, names, and facial features, before processing by AI models.

The platform includes an automated departmental routing mechanism, enabling administrators to specify customized formats for email notifications or API integrations, ensuring rapid and accurate communication with the relevant municipal departments. Finally, to address redundant reporting, the platform leverages a multimodal LLM to detect and group similar issues, effectively reducing administrative workload and streamlining the resolution process.