Flow time: 5 min I your weekly pulse on AI news, tool and case studies reshaping the water sector
🔍 What’s in today’s flow
📊 The World Economic Forum warns that 54% of executives expect AI to displace jobs by 2030, urging water utilities to invest now in workforce training and human-AI collaboration strategies.
💧 UMass Amherst's AI-powered satellite system now monitors water quality and quantity in all rivers wider than 50 meters globally, giving utilities free access to data that previously required expensive field measurements.
⚙️ Gmail's new Gemini AI answers questions buried in years of emails using natural language, potentially saving water professionals hours of searching through project correspondence and vendor communications.
⚠️ Anthropic blocked rival AI companies from using its Claude models, highlighting vendor lock-in risks that utilities should consider when selecting AI partners for critical operations.
🤖Latest in AI: Four futures for AI and jobs by 2030
The details
The World Economic Forum released its report outlining four potential scenarios for how AI and workforce readiness could reshape jobs by 2030. The scenarios range from Supercharged Progress, where exponential AI advancement meets widespread workforce readiness enabling an agentic leap in productivity, to Age of Displacement, where rapid AI progress outpaces workforce adaptation, fueling mass automation and inequality. Between these extremes lie Co-Pilot Economy, characterized by gradual AI integration with human collaboration, and Stalled Progress, where AI development slows amid talent shortages and infrastructure constraints. The report found that 54.3% of executives globally expect AI to displace existing jobs, while only 23.5% anticipate net job creation
Why it matters
Water utilities face similar workforce transformation pressures as other industries but with unique operational constraints. The scenarios suggest utilities should prepare for multiple futures simultaneously by investing in human-AI collaboration for tasks like network monitoring and treatment optimization while building workforce adaptability through continuous training programs. For rural and smaller utilities with aging workforces and recruitment challenges, AI tools could help bridge knowledge gaps during generational transitions, capturing institutional memory and supporting less experienced operators. However, the displacement risks are real. Utilities must balance automation benefits against maintaining skilled staff for emergency response, regulatory compliance, and community trust, where human judgment remains irreplaceable.
🔬AI research spotlight: AI transforms wastewater treatment operations
Source: Sciencedirect
Researchers have demonstrated how artificial intelligence and digital twin technology are revolutionizing wastewater treatment plants (WWTPs) by creating virtual replicas that simulate and optimize processes. The study reviewed machine learning, deep learning, and hybrid AI models for managing emerging contaminants including PFAS, heavy metals, microplastics, and antibiotics while supporting circular economy principles.
The details
Digital twins enable simulation and optimization of WWTP processes without disrupting real-world operations, allowing testing of treatment scenarios for emerging contaminants
AI models integrate circular economy principles, extracting valuable materials like nutrients and energy from wastewater while reducing operational costs
Critical barriers include incomplete AI integration with existing infrastructure, data quality issues from sensor drift and monitoring gaps, and lack of standardized protocols
Most facilities lack unified frameworks combining forecasting accuracy, operational optimization, and uncertainty management in single treatment processes
Why it matters
This research addresses urgent needs for rural and small utilities struggling with emerging contaminants and operational costs. AI-driven digital twins offer affordable pathways to test treatment upgrades virtually before capital investments, while predictive models help resource-constrained facilities optimize chemical dosing and energy use. The emphasis on standardized data protocols and explainable AI systems directly supports regulatory compliance and builds operator confidence, making advanced treatment accessible beyond large metropolitan utilities.
🔧 Case study: Satellite AI tracks global river health from space
What happened
University of Massachusetts Amherst researchers developed Confluence, an open-source software framework that combines artificial intelligence with satellite data from NASA's SWOT mission, Landsat, and Sentinel-2 to monitor water quantity and quality in all rivers wider than 50 meters globally. The computer vision algorithm was trained to automatically detect rivers and distinguish between clouds, snow, and terrain shadows without requiring external inputs like elevation maps. This enables simultaneous estimation of river discharge and suspended sediment concentration for the first time at a global scale in near real-time.
Why it matters
Confluence provides water utilities with unprecedented access to river monitoring data that was previously unavailable or required expensive field measurements. The platform supports drought and flood prediction, water resource planning for drinking water and irrigation, infrastructure planning, and environmental monitoring. For rural utilities with limited monitoring budgets, this free satellite-based system offers critical data on upstream conditions, sediment loads affecting treatment costs, and flow patterns for source water protection. The near real-time capability enables integration into operational forecasting models, helping utilities prepare for changing conditions before they impact operation
🔧Trending tool: Gmail with Gemini integration
Source: Nano Banana, Inside AquAI
Gmail's new Gemini-powered features transform email from a static archive into an interactive knowledge base. AI Overviews answer questions using natural language, instantly surfacing information buried in years of correspondence. Help Me Write generates draft emails from scratch, Suggested Replies provide one-click responses matching your tone, and the coming AI Inbox automatically prioritizes messages. Rolling out free to all Gmail users, with premium features for Google AI Pro and Ultra subscribers.
Key features
AI Overviews summarize lengthy threads and answer complex queries like "Find renovation quotes from last year" without keyword searches
Help Me Write and Suggested Replies generate contextual email drafts that match personal writing style, reducing composition time
AI Inbox (in testing) identifies VIPs, highlights to-dos, and surfaces time-sensitive items based on relationship signals and message content
⚖️ AI Tool Scorecard
Ease of use: 5/5 - Natural language interface requires no training; features integrate seamlessly into familiar Gmail interface
Cost: 3/5 - Core features free; premium capabilities (Proofread, advanced queries) require Google AI Pro ($20/month) or Ultra subscription
Security and privacy: 4/5 - Analysis occurs securely with Google's existing privacy protections; user data remains under personal control
Integration: 5/5 - Native Gmail integration with plans to bring context from other Google apps (Calendar, Drive, Docs) for enhanced personalization
Overall:17 /20 - This represents a genuinely useful productivity tool for water professionals managing high email volumes across multiple projects and stakeholders. The ability to query historical correspondence and auto-generate contextual responses could save hours weekly, though organizations should establish guidelines around AI-drafted communications for regulatory or legal matters to ensure appropriate oversight.
🧠 Deep prompt: Master prompt design with Google
Google's prompt design documentation outlines four key components for effective AI interactions. First, clearly define your task through specific questions or instructions. Second, use system instructions to set the model's role, tone, and constraints before any user input. Third, include few-shot examples showing the desired output format and style. Fourth, provide relevant contextual information like tables or background data that inform the response.
Practical guidelines
Write tasks as specific questions or clear instructions, avoiding ambiguity
Use system instructions to establish expertise level and communication style upfront
Provide 2-3 examples demonstrating your preferred format before asking for new outputs
Include relevant data, tables, or background context the model needs to reference
Structure complex prompts with clear sections using markdown or labeled components
Start simple and iterate, adding complexity only when needed
Experiment with parameters like temperature to balance creativity and precision
For safety-sensitive applications, set explicit content boundaries in system instructions
Test prompts with edge cases to identify failure modes before production use
🕵️The shadow of AI: AI vendors block competitors from using their models
Source: Rundown
Anthropic has cut off access to its Claude AI models for competitor xAI, whose engineers had been using Claude through the Cursor coding platform to accelerate their own AI development work. In an internal email, xAI cofounder Tony Wu acknowledged the company would take a productivity hit but framed it as motivation to build in-house tools. This follows similar restrictions Anthropic imposed on OpenAI and the coding service Windsurf.
Why it matters
This escalation highlights growing friction in AI development as companies protect competitive advantages through access controls. For water utilities evaluating AI vendors, it reveals the risks of depending on proprietary systems that can be restricted or withdrawn. Vendor lock-in becomes a technical reality when competitors are systematically blocked from using each other's tools, potentially limiting innovation and raising switching costs for enterprise customers.
Takeaway
Diversify AI tool portfolios and maintain vendor independence to avoid disruption from competitive restrictions between providers.
Thanks for reading! I hope you’ve enjoyed this week’s edition and look forward to seeing you next week!
Dr. Andrea G.T