In partnership with
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
Global AI-agent standards dropped this week, giving utilities a clearer path to safer automation.
Predictive-monitoring rollouts are accelerating as more utilities use electrical signatures to catch pump failures early
A new wave of low-cost, high-power AI models is emerging, opening doors for small councils to run advanced analytics
Cyber risks are rising fast: even Google’s new AI tool was hacked within 24 hours of launch
The Future of Shopping? AI + Actual Humans.
AI has changed how consumers shop by speeding up research. But one thing hasn’t changed: shoppers still trust people more than AI.
Levanta’s new Affiliate 3.0 Consumer Report reveals a major shift in how shoppers blend AI tools with human influence. Consumers use AI to explore options, but when it comes time to buy, they still turn to creators, communities, and real experiences to validate their decisions.
The data shows:
Only 10% of shoppers buy through AI-recommended links
87% discover products through creators, blogs, or communities they trust
Human sources like reviews and creators rank higher in trust than AI recommendations
The most effective brands are combining AI discovery with authentic human influence to drive measurable conversions.
Affiliate marketing isn’t being replaced by AI, it’s being amplified by it.
🔬AI research spotlight: Smarter, Safer AI Agents
Source: World Economic Forum
The details
The World Economic Forum released a new framework for evaluating AI agents used in real-world systems, focusing on reliability, oversight, and safety. The research outlines technical standards, testing methods, and governance principles for high-risk environments.
Key points
• Provides structured tests for agent behaviour under uncertainty.
• Introduces scoring methods for transparency and accountability.
• Defines risk tiers to guide safe deployment.
• Explains how to monitor AI agents over time, not only at launch.
Why it matters
Water utilities deploying AI for planning, optimisation, and operations can use this framework to reduce system risks, improve trust, and ensure automated agents behave predictably, especially for small councils beginning their digital-transformation journey.
🤖Latest in AI: DeepSeek debuts new AI models
Source: deccanchronicle.com
The details
DeepSeek released a new family of high-performance AI models aimed at competing with Google and OpenAI. The models offer strong reasoning performance, faster inference, and lower hardware requirements, making them more accessible for organisations with limited budgets. Early benchmark results show significant gains in efficiency and task accuracy.
Why it matters
Cheaper, faster AI models mean councils, regional water authorities, and small engineering teams can run advanced analytics without expensive cloud setups. This reduces barriers for leak detection, treatment optimisation, asset forecasting, and real-time monitoring - enabling more utilities to use AI to improve safety, cost efficiency, and planning decisions.
🔧 Case study: Southern Water cuts failures with real-time monitoring
Source: thewaternetwork.com
What happened
Southern Water partnered with Samotics to deploy real-time asset-monitoring across its wastewater network. Using electrical-signature analysis, Samotics’ system continuously tracks pump and motor performance, identifying inefficiencies, early faults, and energy-wasting behaviours. The rollout covers critical sites where failures often trigger overflows or emergency responses.
Why it matters
For water utilities, continuous electrical-signature monitoring provides earlier warnings than traditional vibration or SCADA alerts. This helps reduce unplanned outages, lower energy costs, prevent sewer spills, and support compliance with strict overflow limits. It also builds a stronger data foundation for predictive maintenance programs across ageing assets.
🔧Trending tool: Prompt Genie for better prompts
Source: prompt-genie.com
Prompt Genie helps users create clearer, more structured prompts for GPT-style models. It analyses your draft, fixes ambiguity, suggests improvements, and provides templates for technical tasks. For water utilities and consultants, it can speed up standardised reporting, risk reviews, and engineering-workflow automation.
Key features
Converts short, messy instructions into detailed, high-quality
Works across multiple LLM platforms (ChatGPT, Claude, Gemini, Copilot, etc.).
Lets users save, reuse, and share prompt templates for recurring
⚖️ AI Tool Scorecard
Ease of use: ⭐⭐⭐⭐ , simple Chrome extension and web app; quick to learn, though non-technical users may still need examples.
Cost: ⭐⭐⭐⭐, offers a free tier with paid upgrades, making it accessible for small teams and pilot projects
Security & privacy: ⭐⭐⭐, Runs via browser and cloud; fine for non-sensitive prompts, but utilities should avoid pasting confidential operational or customer data.
Integration: ⭐⭐⭐ , Integrates smoothly into browser workflows but not directly into SCADA, asset systems, or GIS; copy–paste is still needed
Overall: 14/20 - Prompt Genie is a strong helper for standardising and improving prompts, boosting quality for automation, reporting, and QA. Its main gap is enterprise integration, but as a low-cost skills accelerator, it suits most water-sector teams.
🕵️AI’s shadows: Google’s new AI coding tool didn’t last 24 hours before being hacked
Source: facebook.com
A high-profile Google research model was accessed by hackers, raising alarms about security gaps in experimental AI systems. The incident showed how quickly sensitive models can leak when protection layers fail, especially during rapid innovation cycles.
Why it matters
Water utilities and public-service operators increasingly rely on AI tools for monitoring, optimisation, and decision support. A breach in one of these systems could expose operational data, compromise safety controls, or disrupt service continuity, highlighting the need for stronger cybersecurity and governance.
Takeaway
AI adoption must grow alongside security, not ahead of it. Protecting critical-infrastructure data is non-negotiable.
🧠 Deep Prompt Dive – how to write better prompts that sound like humans
AI can help with water operations, reporting, and treatment optimisation — but the output often sounds like an AI wrote it. Operators, engineers, and regulators need clear, practical language that matches how people actually speak and write onsite. This prompt solves that by rewriting text into a direct, human style while keeping all technical meaning accurate.
Try this prompt: Act as my Plain-Language Editor for water professionals. When I paste text, rewrite it so it is clear, direct, and easy to read, like something a competent engineer or operator would write - not like an AI model or generic tech blogger.
Rules
Keep the original meaning; only change the tone to sound human and practical.
Preserve all ideas, claims, point of view, tense, and technical details.
Keep a similar length and structure unless it’s a banned style.
Don’t add examples, assumptions, numbers, or new information.
Output only the revised text — no notes or explanations.
Replace banned words with simple, concrete language.
Avoid AI-style language, buzzwords, or corporate phrases.
Use straightforward terms familiar to operators and engineers.
Keep the tone concise, direct, and suitable for operational communication.
Thanks for reading! I hope you’ve enjoyed this week’s edition and look forward to seeing you next week!
Dr. Andrea G.T