Detailed description of project
AGORA-AI is an AI-powered decision-support system designed to help municipalities and public authorities make evidence-based decisions when addressing complex urban challenges.
Developed by DOTSOFT S.A., the solution builds upon the NOON-IoT smart-city digital-twin platform, combining real-time operational data with scientific evidence through Knowledge Graphs, Large Language Models (LLMs), and Retrieval-Augmented Generation (RAG).
AGORA-AI enables municipal administrators to describe urban problems in natural language and receive ranked intervention recommendations based on comparable cases from other cities. Each recommendation includes projected impact ranges, explicit uncertainty estimates, and traceable scientific references.
The system also supports what-if scenario simulations, allowing public authorities to evaluate alternative interventions before implementation. Its initial areas of application include urban mobility and environmental management, with the potential to expand into additional smart-city domains.
Type and scope of work provided
- Knowledge Graph Enhancement: Extending the existing knowledge graph to connect urban problems, interventions, city characteristics, and measurable outcomes.
- AI Integration: Integrating European Large Language Models and embedding/vector services through the AI-on-Demand (AIoD) Platform.
- Similarity and Scenario Analysis: Enhancing hybrid semantic and graph-based similarity mechanisms to identify comparable urban cases and estimate intervention impacts.
- Multilingual Capabilities: Expanding the natural-language interface to support multiple European languages.
- Interoperability and Data Integration: Connecting the solution with NOON-IoT and standards-compliant smart-city infrastructures through NGSI-LD and FIWARE.
- Pilot Deployment and Validation: Deploying and evaluating AGORA-AI with a public-sector AI Adopter using operational municipal data.
AGORA-AI aims to strengthen transparent, accountable, and sustainable urban governance by transforming operational data and documented evidence into actionable recommendations, while ensuring that final decisions remain under human control.
The project is proposed within the AI-on-Demand (AIoD) / DeployAI programme, supporting the adoption of trustworthy European AI technologies in public administration.
