Droplet

Droplet: real-time urban water network intelligence. Fusing satellite & hydraulic data to optimize distribution.

  • 0 Raised
  • 1,214 Views
  • 0 Judges

Tags

  • Italy

Categories

  • Challenge #1: Securing equitable and efficient access to water ​

Gallery

Description

Droplet: AI-Powered Water-Network Intelligence

Droplet is an AI-powered water-network intelligence platform designed to secure more equitable and efficient access to water. It helps utilities reduce water losses, detect network stress, and prioritize interventions before small anomalies become major failures.

The problem we address is Non-Revenue Water. In many European territories, a large share of treated water is lost before reaching citizens because of leaks, pressure issues, aging infrastructure, limited telemetry coverage, and slow incident localization. This creates higher operating costs, inefficient use of water resources, and unequal service quality for communities.

Our pilot area is Ciociaria, in the province of Frosinone, where water losses reach 69.5% according to ISTAT data. This makes it a strong real-world case for testing a scalable solution in a territory where efficient water management is an urgent operational need.

The Digital Twin of Water Infrastructure

Droplet combines satellite intelligence, hydraulic simulation, network topology, and regulatory knowledge into one operational platform. By integrating these diverse data streams, Droplet functions as a digital twin of the water network, providing a real-time virtual representation that mirrors the physical infrastructure's behavior.

Within this digital twin environment, the platform:

  • Produces a Pipe Health Index for each network segment.

  • Highlights high-risk areas on an interactive map.

  • Monitors tanks and District Metered Areas (DMAs).

  • Utilizes an auditable AI agent to explain why an anomaly is relevant and propose actions for human review.

The platform does not replace utility operators or existing SCADA systems. It works as an intelligence layer that helps utilities act faster, justify maintenance decisions, and allocate resources where they can have the highest impact on water conservation and service reliability.

Project Link: https://droplet-alealdini2007-6759s-projects.vercel.app

  • Username: operator

  • Password: droplet-2026

The Breakthrough: Multimodal Semantic Routing Agentic Graph RAG

Droplet represents a paradigm shift in industrial AI. We have pioneered the use of a Multimodal Semantic Routing Agentic Graph RAG, which is the first of its kind used to manage complex utility data.

Advantages over Traditional RAG:

  • Contextual Reasoning: By mapping the network as a knowledge graph, our AI agents understand the physical and logical relationships between pipes, sensors, and environmental data. It doesn't just search; it reasons about system health.

  • Multimodal Semantic Routing (BRICK): Developed by Francesco Massa, the BRICK technology acts as the platform's cognitive traffic controller. It assesses the difficulty and modality (images, documents, audio, or telemetry) of every query and automatically routes it to the most performance-efficient Regolo model.

  • Computational Efficiency: By matching the model size to the query complexity, we drastically reduce token consumption and latency, making high-end intelligence scalable for daily utility operations.

  • Auditable Evidence: The system provides transparent, source-backed explanations for every decision, ensuring human operators can verify every maintenance recommendation.

Data Foundation: Earth & Space Intelligence

Droplet uses advanced remote sensing and signals as a core part of the solution. We have moved beyond legacy systems by integrating NASA data and Cassini-mission-inspired analytical techniques.

  • NASA Data & Cassini Techniques: We utilize high-resolution satellite data to monitor environmental signals. We apply processing techniques derived from the Cassini mission to filter background noise from complex environmental data, allowing us to isolate subtle physical signatures of infrastructure stress.

  • Copernicus Data: Sentinel-2 supports vegetation and moisture anomaly detection around pipe corridors. Sentinel-3 contributes land surface temperature patterns. ERA5 climate data helps distinguish weather-driven effects from infrastructure-driven anomalies. EGMS ground-motion data identifies subsidence and terrain instability. JRC Global Surface Water adds seasonal context.

  • Complementary Datasets: We integrate ECOSTRESS for higher-resolution thermal insights and GRACE-FO for aquifer-level context.

Sustainable Infrastructure: Regolo & Green Energy

A core pillar of our mission is environmental responsibility. Droplet’s AI is powered by Regolo, a cloud inference provider that is completely powered by green energy.

Our inference runs on Seeweb data centers, which operate on 100% renewable energy. We use their models for this hackathon not just for their technical superiority, but because they share our vision of a sustainable future. By optimizing our workflows through BRICK, we ensure that our carbon footprint remains minimal while our diagnostic impact is maximized.

EU Space for Water

Droplet addresses Challenge #1: Securing equitable and efficient access to water.

Efficient access is about protecting the water that is already treated and distributed. Droplet identifies where losses are likely happening by combining satellite signals, hydraulic behavior, and network topology. It improves maintenance planning by ranking segments by operational risk, reducing response time, and ensuring limited resources are prioritized where they have the highest impact.

The Team

  • Alessandro Aldini (Solution Architect & Team Leader): Works on AI systems, product architecture, and automation. NASA Space Apps Rome winner (2023, 2024), creator of ZeroHR, and technology partner to Acea.

  • Francesco Massa (Infrastructure & Design): R&D at Seeweb, creator of the BRICK architecture. Specialist in AI infrastructure, ontologies, and production-grade agent workflows.

  • Sergio Aldini (Marketing Consultant): Strategic investment, business positioning, and public-sector innovation expert with a focus on AI applications.

Conclusion

Droplet turns space data and complex infrastructure signals into operational intelligence. We offer a practical path toward saving treated water, protecting communities, and making every intervention more targeted, transparent, and impactful. Our vision is simple: the water we already have should not be lost in the dark. Droplet gives utilities the visibility to find it, protect it, and deliver it where it matters.


DISCLAIMER for the youtube video: Romolo Mairelli is not part of team and there is Sergio Aldini

Attachments