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
🌾 Kilimo's AI-powered water management platform saved 74 million cubic feet of water across Chilean farms, combining satellite data, weather forecasting, and field monitoring.
🤖 Claude Opus 4.5 now delivers state-of-the-art coding and agentic capabilities at $5/$25 per million tokens, making advanced AI accessible to more water utilities.
⚠️ Research shows AI-written papers with complex language were less likely to be published in journals, revealing quality concerns despite productivity gains.
🔧 Context engineering is replacing prompt engineering as the critical skill for AI deployment, focusing on what information models access rather than how questions are asked.
🛠️ Microsoft developed an agentic AI monitoring system using Azure OpenAI to track water savings in real time, enabling scalable water conservation projects.
This newsletter you couldn’t wait to open? It runs on beehiiv — the absolute best platform for email newsletters.
Our editor makes your content look like Picasso in the inbox. Your website? Beautiful and ready to capture subscribers on day one.
And when it’s time to monetize, you don’t need to duct-tape a dozen tools together. Paid subscriptions, referrals, and a (super easy-to-use) global ad network — it’s all built in.
beehiiv isn’t just the best choice. It’s the only choice that makes sense.
🔬AI research spotlight: AI and gamification drive water savings in drought-stressed basin
Researchers developed the Smart Gamified Water Conservation System (SGWCS), combining Internet of Things (IoT) sensors, a CNN-Attention-LSTM forecasting model, and gamification strategies across residential, industrial, and municipal sites in Raipur, India.
The details
The CNN-Attention-LSTM model achieved 97.2% accuracy in real-time water demand forecasting
Hybrid rule-machine learning anomaly detection reached 92.8% sensitivity while reducing false positives by 38%
Gamification with AI-personalized nudges increased user participation by 28% and reduced residential water use by 12.5%
The system deployed privacy-by-design architecture compliant with GDPR and CCPA standards
Why it matters
This advancement shows how AI can engage communities while optimizing water use, offering rural utilities and water-stressed regions a proven model for combining behavioral science with predictive analytics to achieve measurable conservation outcomes.
🔧 Case study: Chilean farms cut water use 40% with AI-driven irrigation
Source: microsoft
What happened
Kilimo, a water stewardship platform operating across seven countries, deployed an AI monitoring system in Chile's Maipo River basin involving 11 farms. The platform combines satellite imagery captured every 60 hours, meteorological forecasting, soil analysis, and evapotranspiration calculations to issue weekly irrigation recommendations. Microsoft Azure Kubernetes Service powers the real-time calculations, while three specialized AI agents manage project tracking, contract data, and reporting.
Why it matters
This approach demonstrates how utilities can scale water conservation through verified, auditable results. Farmers reduced irrigation by 15% to 20% on average, with some achieving 40% reductions. The success attracted over $25 million in new funding to expand water restoration across 15 watersheds, proving that data-driven water management can secure financial support while delivering measurable environmental benefits.
🤖Latest in AI: Claude Opus 4.5 delivers breakthrough coding and agentic performance
Source: claude.AI
The details
Anthropic released Claude Opus 4.5, achieving state-of-the-art performance on software engineering benchmarks including SWE-bench Verified. The model introduces an effort parameter allowing developers to balance speed and capability, using up to 76% fewer output tokens while maintaining performance. Pricing dropped to $5/$25 per million tokens, making advanced AI accessible to more organizations.
Why it matters
Water utilities managing complex infrastructure can now deploy AI agents for long-running tasks such as asset management, compliance reporting, and system optimization at significantly lower cost. The model's improved reliability and resistance to prompt injection attacks addresses critical security concerns for utilities handling sensitive operational data, while its enhanced coding capabilities can accelerate digital transformation initiatives.
🔧Trending tool: Praxie, no-code AI platform for business operations
Praxie is an AI-powered business operations platform that enables users to create custom applications, dashboards, and automated workflows through conversational interfaces without coding. The platform features vibe coding (AI-generated applications), multi-agent workflow automation, and universal data integration. Trending for its ability to rapidly digitize business processes, Praxie offers pre-configured solutions for operations, quality management, supply chain, and asset tracking relevant to water utilities.
Key features
AI-powered dashboard builder that converts spreadsheet data into visual analytics and recommendations
Multi-agent automation system for handling repetitive tasks including compliance audits and reporting
Universal data integration supporting REST APIs, SQL databases, webhooks, and machine MQTT protocols
⚖️ AI Tool Scorecard
Ease of use: the conversational interface simplifies application creation, though complex workflows may require learning the platform's automation logic⭐⭐⭐⭐
Cost: Pricing information not publicly detailed, typical of enterprise platforms requiring consultation for custom deployments.⭐⭐⭐
Security & privacy: Cloud-based platform with standard enterprise security, but limited public documentation on water sector-specific compliance certifications⭐⭐⭐
Integration: Strong connectivity options including APIs, SQL, and industrial protocols (MQTT) suitable for water infrastructure systems.⭐⭐⭐⭐
Overall: 13/20 -Praxie offers utilities a rapid path to digitization without extensive IT resources. The platform's strength lies in quickly deploying dashboards and automating manual processes, making it valuable for utilities seeking operational efficiency gains. However, organizations should verify security compliance and assess pricing against specific use cases before committing.
🕵️AI’s shadows: AI slop undermines scientific integrity
Source: themarketer.com
A UC Berkeley and Cornell study analysing over one million preprint abstracts found that AI adoption dramatically increased research output (36% to 90% depending on platform), particularly among non-native English speakers. However, AI-written papers using complex language were less likely to be published in peer-reviewed journals, suggesting authors used sophisticated wording to mask weak scientific contributions.
Why it matters
For water research and utilities relying on published studies for decision-making, this trend threatens the integrity of technical literature. The proliferation of AI-generated content with polished language but questionable methodology creates risks for utilities adopting technologies or practices based on flawed research. The water sector depends on rigorous peer review and validated findings to inform infrastructure investments and regulatory compliance.
Takeaway
Complex language no longer reliably signals quality. Critical evaluation of methodology and evidence remains essential when assessing research.
🧠 Deep prompt dive: From prompt engineering to context engineering
The AI field is shifting from prompt engineering (crafting perfect instructions) to context engineering (designing what information AI systems access). As context windows expanded from 4,000 to over one million tokens, the focus moved from optimizing how we ask questions to optimizing what the AI knows when answering.
What context engineering involves
Context engineering means structuring the complete information environment surrounding an AI query. This includes system prompts defining model behavior, relevant data retrieved dynamically, conversation history, available tools, and examples demonstrating desired outputs. The goal is providing exactly the right information at the right time, rather than relying on clever wording alone.
Why the shift matters
Prompt engineering focused on single interactions, while context engineering builds systems that perform consistently across sessions and users. Production AI applications require memory, tool access, and structured data flows that simple prompts cannot provide. Context engineering addresses failures caused by missing information, not inadequate instructions.
Thanks for reading! I hope you’ve enjoyed this week’s edition and look forward to seeing you next week!
Dr. Andrea G.T