AI-Assisted Voice Enabled Computing Framework for Hydrological Analysis

Carlos Erazo Ramirez, Ibrahim Demir

This work presents a web-based, voice-enabled, no-code platform for AI-assisted hydrological analysis. The system allows users to interact through natural language—via both text and speech—to retrieve data, utilize hydrological functions, and visualize spatial and analytical outputs. Core components include a conversational AI assistant utilizing Large Language Models, a modular analysis engine based on HydroSuite, and direct integration with hydrological data from federal agencies using HydroShare and other data and web services. Structured intent parsing, persistent session state, and dynamic map-layer control support multi-turn interactions and reproducible workflows. A case study over the Mississippi River Delta demonstrates how the platform enables guided exploration, layered data integration, and low-latency execution with minimal technical overhead. The platform lowers barriers for research, education, and decision-making in hydrology by combining AI reasoning with a transparent, accessible user interface. By enabling natural language interaction, data integration, and reproducible, multi-turn task processing, this system lays the foundation for automated hydrological research and operational workflows.