Challenges
SmartAqua d.o.o. is a technology company that develops innovative solutions for digitalisation in the utilities sector, with a particular focus on water supply and infrastructure management. In collaboration with several water companies, the SmartAqua team identified common and serious problems in the day-to-day operations of operational teams – from uneven failure records and difficult reporting, to a lack of analytical tools for predicting and optimising field interventions.
With average technical losses in Croatia's water supply systems often exceeding 50%, the need for modernisation is becoming more pronounced. The SmartAqua team has recognized the opportunity to introduce an intelligent system that will ensure better operational efficiency, reduce costs, speed up the reaction to failures and enable a more transparent relationship with end-users.
Solutions
Through the service Test Before Invest offered by EDIH Adria, the SmartAqua team had the opportunity to test a demonstration prototype of the next generation of their platform: SmartAqua WISDOM (Water Intelligent System for Digital Operations Management). The goal was to design a digital solution that combines data from existing systems (GIS data, IoT sensors, user bases) with advanced AI functionalities.
Key components of the tested solution include:
- More MCP AI agents based on GPT-4 Large Language Model (LLM), for:
- Analysis of user fault reports
- Classification and prioritisation of interventions
- generating suggestions for reports and communication with users
- RAG architecture (Retrieval-Augmented Generation) that combines data from a database with generative intelligence
- Visualisation interface with interactive maps, street breakdown statistics, clusters and time intervals
- Advanced dashboards associated with geolocation data, dynamic filters and real-time graphics
- Fault repair module, in which the system recommends the necessary works and equipment, generates reports and monitors the status of field teams
In addition, users without technical pre-knowledge are able to get recommendations, information and even generated reports by simply entering a text description of a conversational-style malfunction – thanks to the power of language models.
Results and Benefits
Solution testing has shown strong transformative potential for SmartAqua and its utilities partners.
Key benefits include:
- Extreme reduction in response time – AI agent responds to queries from users and technical staff within seconds
- Introduction of multi-level automation – including natural language processing, summary creation and spatial data analysis
- More than 300% Increased user interactions during the test period, thanks to an intuitive and accessible interface
- Mapping of critical fault points is enabled automatically, which facilitates investment planning and resolution
- Insight into seasonal trends of failures and frequency in individual geographical zones, which was previously almost impossible to follow
Thanks to the innovative integration of language models with geographic and infrastructural data, Smart Aqua WISDOM enables a whole new way of working – where the system is not only a passive display of information, but an active participant in operational and strategic processes.
Lessons learned
The successful implementation of the pilot project has also brought a number of useful insights that will shape the future development of the solution:
- AI tools can only be effective if they are linked to a clear and structured database – fault, infrastructure and location data must be accurate and standardised
- Local contextualized RAG bases (in Croatian) significantly improve the accuracy of the AI agent's response
- The democratization of analytics through the natural language interface allows all users (not only engineers) to see data and make decisions
- The technological multi-layer system must be scalable – the ability to operate the system with more users and more failures is essential for long-term success
- Collaboration with end-users and field teams is crucial – validation of functionality through concrete scenarios has significantly increased the value of the prototype
Measurable data
Based on the testing, concrete, measurable effects that the implementation would bring have been identified:
- Up to 90% faster answers to complex issues related to fault management
- Automated processing of more than 100 failures per week, with minimal human work
- 300% more user interactions thanks to a custom AI interface
- Savings of 4-6 hours per week per operator, which would otherwise have to review and report status
- Digital assistants generate reports ready for PDF export and e-mail communication to users or administration
Conclusion
SmartAqua d.o.o. has raised the bar in the digitization of utilities through this TBI project. The developed system not only optimizes operational processes and communication, but also uses the power of artificial intelligence to enable predictive analysis and real-time decision-making.
SmartAqua WISDOM Model for the future development of digital companies in the water services sector, and the platform designed during testing can serve as a basis for the implementation of similar solutions in utilities across Croatia and the region.
The next steps include the finalisation of the functional model, its integration with ERP systems and GIS partner bases, as well as preparation for commercial application.
English 
Croatian 
Italian