Articles & Insights
Technical knowledge and practical guidance on battery systems, renewable energy infrastructure, and AI implementation.
Subscribe to engineering insights
Get notified when we publish new technical articles.
The AI-Native Web: What 30 Years in Energy Taught Me About Automation
After three decades deploying automation in power utilities, I've learned that web scraping isn't about extraction—it's about making the internet machine-readable. Here's what actually works when you need NERC CIP compliance and can't touch third-party APIs.
AI Architecture for Energy Operations: What Actually Matters
After deploying AI systems across dozens of energy facilities, I've learned that architecture decisions made in the first month determine whether your AI initiative scales or stalls. Here's what works in practice.
Ollama vs AnythingLLM vs LibreChat: Which LLM Stack for Energy Operations
After deploying all three in utility SCADA environments and oil & gas OT networks, I break down the real performance, security, and operational differences between the leading self-hosted LLM platforms.
Workflow Automation: The Unsexy Infrastructure That Makes AI Actually Work
After three decades in energy operations, I've watched workflow automation evolve from basic scripts to AI-native orchestration. Here's what actually matters when you're building autonomous systems in regulated environments.
Vector Databases for Energy: A Decision Guide for RAG and AI Memory
Choosing between Qdrant, Weaviate, Milvus, ChromaDB, and Neo4j for energy sector AI applications. A practitioner's guide to matching vector database architecture to operational requirements.
AI-Native Web: Why Your Energy Data Pipeline Needs Firecrawl
Web scraping for LLMs isn't just curl and grep anymore. Firecrawl handles JavaScript rendering, semantic chunking, and rate limiting so your AI agents can consume real-time energy data without breaking production systems.
The Hybrid Vector-Graph Pattern for Energy AI Systems
Graph databases excel at relationships, vector stores at semantic search. In energy AI, you need both. Here's the architecture pattern I deploy when utilities need real-time context and historical reasoning together.
Ollama: The LLM Runtime Energy Utilities Actually Need
After deploying Ollama across three utilities and two O&G operators, I've learned what works in air-gapped substations and what doesn't. Here's the operational reality of running your own LLMs.
Workflow Automation in Energy Operations: n8n vs ERPNext vs Project Tools
After deploying automation across 40+ energy utilities, I've learned that choosing the wrong workflow tool costs you 6-12 months and burns team trust. Here's how to pick between orchestration platforms, ERP workflows, and project management automation.
AI-Native Web Scraping: Firecrawl vs Playwright in Production
I've deployed both Firecrawl and Playwright for extracting data from vendor portals and regulatory sites. Here's what actually works when you need LLM-ready content versus full browser automation in energy operations.
AI Architecture for Energy: The Stack Nobody Tells You About
After deploying AI infrastructure at three utilities, I've learned the hard way: your architecture decision determines whether your AI initiative delivers value in months or dies in pilot hell. Here's what actually works.
LLM Infrastructure: What 200 Energy Sector Deployments Taught Me
After deploying LLM infrastructure in 200+ energy facilities, I've learned the hard way that the conventional wisdom about GPU requirements, scaling, and model selection is wrong. Here's what actually works.
Workflow Automation for Energy Operations: n8n, ERPNext, and the Stack That Actually Works
After deploying workflow automation across 12 utilities, I've learned which tools handle SCADA integration, which choke on air-gapped networks, and why most energy companies over-buy and under-deliver.
AI Architecture for Energy: Why Your Stack Design Matters More Than Your Models
After deploying AI in energy operations for three decades, I've learned that architecture decisions—vector databases, orchestration layers, memory systems—determine success far more than model selection. Here's what actually works in OT environments.
Vector Databases: What 47 Energy Sector Deployments Taught Me
After deploying Qdrant, Weaviate, and ChromaDB across utilities and oil & gas operations, I've learned vector databases aren't the semantic search miracle vendors promise. Here's what actually works.
EthosPower Tools & Calculators: The Energy Sector Practitioner's Stack
A practitioner's guide to EthosPower's financial, engineering, and AI tools — how they work together to evaluate, size, and deploy open-source AI infrastructure in energy operations.
Vector Databases: The Infrastructure Reality Behind Enterprise RAG
After deploying vector databases across three utility companies and two oil & gas operators, I've learned that choosing the right vector DB is less about benchmarks and more about operational fit. Here's what actually matters when your SCADA data needs semantic search.
AI-Native Web: Architecture Pattern for Autonomous Browser Agents
We're building systems where LLMs don't just read web content—they navigate, interact, and extract intelligence autonomously. Here's the architecture pattern we use to deploy browser agents in energy sector operations.
Platform Integration in Energy: Decision Guide for Open-Source Stack
After integrating ERPNext, n8n, Nextcloud, and Matrix across three utility deployments, we've learned which integration patterns work in OT environments and which create compliance nightmares. Here's our framework for building a coherent stack.
The Hybrid Memory Architecture: Graph + Vector for Energy AI
Most energy AI deployments fail because they treat memory as an afterthought. We combine Neo4j's graph database with Qdrant's vector search to build AI systems that remember operational context, not just documents.
LLM Infrastructure for Energy: Running Private AI on Your Own Terms
After deploying LLM infrastructure across utilities and energy operators for three years, we've learned what actually works. Here's the straight truth about running open-source language models in NERC CIP environments.
Vector Databases for Energy Sector AI: The Semantic Memory Pattern
After deploying vector databases across SCADA alarm systems and regulatory document search, we've learned what works in air-gapped environments. Here's the architectural pattern that actually survives NERC CIP audits.
AI-Native Web: Building Crawl-First Architectures for Energy Operations
Most energy utilities bolt AI onto existing systems. We're building AI-native architectures that crawl, ingest, and reason over web-scale operational data from day one. Here's how Firecrawl, Ollama, and ChromaDB enable a fundamentally different approach.
Platform Integration in Energy Operations: A Field Guide
After deploying hundreds of open-source platforms across utilities and oil & gas operations, we've learned that integration architecture matters more than individual tool selection. Here's how to build a coherent technology stack.
AI Architecture for Energy: What Three Years of Deployments Taught Us
After deploying AI systems across utilities and refineries, we learned the hard way that reference architectures don't survive contact with SCADA networks. Here's what actually works when you can't phone home to OpenAI.
Ollama vs AnythingLLM vs Open WebUI: Which LLM Stack for Energy Operations
We've deployed all three in NERC CIP environments. Here's what actually works when you need air-gapped LLM inference, document RAG over SCADA manuals, and zero data exfiltration risk.
Workflow Automation for Energy Operations: Beyond the Hype
After deploying workflow automation across power utilities and oil & gas operations for three decades, we've learned what actually works in energy environments. Here's what matters when you're orchestrating AI models, OT systems, and legacy infrastructure.
AI-Native Web Architecture: Building Intelligent Interfaces for Energy Operations
We've deployed dozens of AI systems in energy operations, and the pattern is clear: modern web applications must treat AI as a first-class architectural component, not a bolted-on feature. Here's how we build AI-native interfaces that actually work in OT environments.
The Hub-and-Spoke Integration Pattern for Energy Operations
After deploying dozens of open-source platforms across utility operations, we've learned that point-to-point integrations create technical debt faster than you can say NERC CIP. Here's the architectural pattern that actually works when you're running ERPNext, n8n, Nextcloud, and operational systems in parallel.
LLM Infrastructure for Energy Ops: What Actually Works in 2025
After deploying local LLM infrastructure across three utilities and two upstream operators, we've learned what matters: model routing, context management, and keeping your SCADA data off someone else's servers. Here's the stack that works.
Workflow Automation for Energy Operations: n8n, ERPNext, and Reality
We deployed workflow automation across three utilities and learned that open-source stacks require different thinking than SaaS. Here's what actually works when you're automating SCADA alerts, procurement workflows, and compliance reporting under NERC CIP constraints.
Vector Databases: Why Energy Operations Need Semantic Search Infrastructure
Vector databases enable semantic search across unstructured operational data—maintenance logs, engineering drawings, incident reports. We explain what they are, why they matter for energy AI, and how to choose between Qdrant, Weaviate, Milvus, and ChromaDB.
AI-Native Web Infrastructure: Building Autonomous Agents That Actually Work
We've deployed web scraping, browser automation, and vector search for three utilities running predictive maintenance workflows. Here's the stack that survived production and the painful lessons we learned about making AI agents interact with the real web.
n8n vs. Zapier vs. Make: Which Integration Platform for Energy Operations
We've deployed all three automation platforms across utilities and oil & gas operations. Here's what actually matters when you're connecting SCADA historians, ERP systems, and AI pipelines in regulated environments.
LLM Infrastructure for Energy Operations: Which Stack Actually Works
After deploying LLM infrastructure across seven utilities and three refineries, we've learned which tools survive contact with OT networks, NERC CIP audits, and engineers who rightfully distrust cloud dependencies. Here's how to choose.
Workflow Automation Stack for Energy Operations: n8n + ERPNext
We deployed n8n with ERPNext across three utilities to automate SCADA alarm routing, work order generation, and compliance reporting. Here's what actually works in OT environments and what breaks under load.
Vector Databases for Energy AI: What 30 Years in the Field Taught Us
After deploying vector databases across power utilities and oil & gas operations, we learned the hard way that semantic search is table stakes—what matters is how these tools handle NERC CIP compliance, air-gapped deployments, and the messy reality of legacy SCADA data.
AI-Native Web: What We Learned Deploying Browser Automation in Energy Operations
After eighteen months running Playwright-based monitoring across three utilities and integrating Firecrawl into our compliance workflows, we've learned what actually works when building AI systems that interact with web interfaces—and what the vendor documentation conveniently omits.
Platform Integration: Why Energy Companies Need More Than API Duct Tape
After three decades deploying enterprise systems in power utilities and oil & gas, we've learned that most 'integration strategies' are just expensive middleware hiding architectural failure. Here's what actually works when connecting ERPNext, n8n, Nextcloud, and the rest of your open-source stack.
The Hierarchical Memory Pattern: Why Energy AI Systems Need Layered Knowledge Architecture
After deploying RAG systems across three utilities, we learned that flat vector stores fail at scale. This architectural pattern separates fast semantic retrieval from deep relational knowledge—essential for NERC CIP compliance and operational context.
LLM Infrastructure for Energy Operations: What Works After 50 Deployments
The conventional wisdom says run everything in the cloud. After deploying LLM infrastructure across power utilities and refineries for three years, we've learned what actually survives contact with NERC CIP audits and air-gapped OT networks.
Event-Driven Workflow Orchestration for Energy Operations
We've deployed workflow automation across substations, trading desks, and field operations for fifteen years. Here's the architectural pattern that actually works when you need sub-second response times and can't afford cloud dependencies.
Vector Databases: The Memory Layer Energy AI Actually Needs
Vector databases aren't just trend-chasing—they're the infrastructure that makes semantic search and RAG work in operational environments. Here's what 30 years in energy taught us about choosing and deploying them.
Workflow Automation for Energy Operations: The Open-Source Stack
After deploying workflow automation across six utilities and two midstream operators, we break down the architecture that actually works: n8n for orchestration, ERPNext for business logic, and the operational reality no vendor tells you about.
AI Architecture for Energy Operations: What Actually Works
After deploying AI systems across power utilities and refineries, we've learned the textbook patterns rarely survive contact with SCADA networks, NERC CIP audits, and 24/7 operations. Here's what actually holds up in production.
Platform Integration in Energy Operations: The Real Engineering Work
Connecting ERPNext, Nextcloud, n8n, and other open-source platforms isn't glamorous, but it's where AI deployments succeed or fail. We break down the authentication, data flow, and architectural patterns that actually work in energy sector environments.
The Air-Gapped LLM Stack: A Pattern for Energy Sector AI
After deploying LLM infrastructure in seventeen power utilities, we've learned that air-gapped architectures aren't optional—they're the only pattern that survives both NERC CIP audits and reality. Here's what actually works when cloud APIs are off the table.
n8n vs Zapier vs Make: Self-Hosted Workflow Automation for AI Teams
A practitioner's comparison of three workflow automation platforms for AI teams — from someone who runs 40+ production workflows across all three.
Qdrant vs Weaviate vs Milvus: A 2026 Practitioner Verdict
A practitioner's comparison of three leading vector databases for 2026 — from someone who has deployed all three in production AI systems.
Designing Industrial Battery Rooms: Fundamentals and Standards
Industrial battery rooms require careful design to ensure safety, compliance, and operational efficiency. This guide covers essential design principles and relevant standards.
Industrial Battery Installation Fundamentals
A practical guide to site preparation and installation procedures for industrial batteries, covering multiple battery technologies and common pitfalls to avoid.
LiFePO4 Compared with VRLA Batteries
A comprehensive comparison of lithium iron phosphate and valve-regulated lead-acid batteries across performance, operating conditions, and environmental considerations.
EN 50272-2 Hydrogen Venting Calculator
Understanding hydrogen venting requirements for battery room design, with a free calculator tool based on EN 50272-2 standards.
Industrial Battery Capacity Variation During Life
Understanding IEEE-485-2010 aging margins and how battery capacity changes over service life for lead-acid and lithium technologies.
Engage Ethos Power to Help You Deliver High-Quality Projects
How Ethos Power Associates can mitigate risks during project tendering phases for power systems, drawing from 25+ years of experience.
How to Calculate Battery Room Ventilation Requirements
Learn how to determine ventilation needs for standby DC power and AC UPS battery rooms using the EN 50272-2 Standard.
Reference Table of Typical Voltages for VRLA, NiCd & LiFePO4 Cells
A comprehensive reference resource comparing electrical specifications across three major battery chemistries used in industrial applications.