New Research and Deals Bring AI to the Water Table

New MARLIN research and moves by Intelligent Core and SePRO highlight early momentum in AI enabled water management

4 Dec 2025

Artificial intelligence is drawing increasing attention in the US water sector, driven by new academic research and early commercial moves. While adoption remains limited, recent developments suggest a rising interest in using machine learning and analytics to improve water management.

A key example is MARLIN, a reservoir coordination framework introduced as a preprint in September 2025. The system integrates multi-agent reinforcement learning with language model based reasoning. Early simulations using historical US Geological Survey data showed stronger performance in handling uncertainty and faster modeled responses to potential flooding. Although still confined to research settings, MARLIN has prompted discussion about how AI could support decision-making during volatile hydrologic events.

Commercial activity is emerging alongside the research. Intelligent Core has extended its AI platform, first built for produced water operations in oil and gas, to explore broader water-related uses. SePRO has moved further, acquiring Arietta.ai on August 26 and launching SePRO Digital in early September. The company cited rising environmental standards and tightening regulation as catalysts for its expansion into software solutions.

Analysts caution that progress will be uneven. Many utilities face fragmented data systems, inconsistent monitoring networks and legacy infrastructure that complicate integration with new digital tools. Slow regulatory cycles and limited funding, especially for smaller utilities, remain persistent barriers to large-scale implementation.

Even so, the combination of new research, targeted investment and mounting climate pressures is beginning to reshape the sector’s outlook. AI tools could eventually help utilities strengthen situational awareness, optimise water use and enhance long-term planning.

As water systems contend with increasing extremes, from sudden floods to extended droughts, interest is expected to remain high in technologies capable of synthesising complex data and identifying emerging risks. The next phase will test which AI concepts can move from experimental modelling to practical deployment, and whether digital capabilities can meaningfully improve the resilience of US water infrastructure.