John Holliday

From the Preface, for those wondering about AI Powered Digital Twins #newbook Wiley 2026 AI-powered digital twins are the culmination of a century of scientific and digital progress—a living mirror of physical assets, systems, and processes that ingests real-time data, runs AI-driven simulations, and optimizes performance. They unify fragmented technologies into one adaptive system where physical and digital worlds continuously inform and strengthen each other. By combining telemetry, diverse data, regulations, and operational constraints, a twin preserves structural integrity while running rapid what-if scenarios. AI makes these systems dynamic and scalable, reducing inefficiency and strengthening security through real-time detection and resilience planning. I am fond of enterprise and business modeling, supply chain and logistics, and securing our critical infrastructure, particularly the energy sector, thus, many of our real world examples detail those. This book has four recurring themes: * Human-Centered Culture, Strategy, and Business AI must enhance—not replace—human capability. Innovation cultures, cross-sector literacy, and systems thinking allow humans to guide the systems they build. * Engineering Reality: Architecture, Infrastructure, and Data Foundations We show how AI integrates with OT, IT, energy systems, supply chains, and workflows. Knowledge graphs, hierarchical tagging, and multimodal data anchor trustworthy digital twins. * AI Agents, Automation, and Execution AI agents automate domain workflows while keeping humans in the loop. Modular architectures and clear governance ensure adaptability, accountability, and measurable value. * Security, Safety, and Defensive Digital Twins AI expands attack surfaces and IT and OT security gaps persist. Digital twins become defensive systems—continuously monitoring, simulating, and protecting critical infrastructure with engineering rigor. The book has four parts, with 17 self-contained chapters that you can read in any order: * Part 1: AI-Native Digital Twins * Part 2: Infrastructure and Security Requirements * Part 3: Engineering * Part 4: Maintenance, Sustainment, and Business The sections oscillate in and out of technical rigor and difficulty, so pick and choose according to your own needs and at your leisure. Photo today: Military Order of the Caribao Wallow- for those who appreciate 🇺🇸

28

9 Comments

AI ethics starts below the surface. We talk a lot about the "fairness" of models. The bias in data. The transparency of outputs. But here’s a harder question: What GOOD is an ethical model trained on extractive infrastructure? If the system powering your AI consumes millions of gallons of water, runs on coal heavy grids, and concentrates compute access in the hands of a few..... Is the model still ethical? I don’t think ethics ends at the dataset. It begins with the land we build on. The power we draw from. The people we include or don’t. AI isn’t abstract. It has weight. It has temperature. It has cost. And we have to get honest about the systems underneath it. Curious to hear your thoughts: Where do you think the ethical line in AI really starts? #ArtificialIntelligence #datacenters #aiethics #GB200

3

🚨 Meta’s AI agent just triggered a Sev-1. Here’s why guardrails for data is so critical in the Agentic AI era. The Incident & Architectural Failure Last week, an internal Meta AI agent autonomously posted flawed technical instructions. When executed by a human engineer, it bypassed standard IAM controls, triggering a chain reaction that exposed proprietary code and sensitive user datasets for two hours. The core issue wasn't just a hallucination; it was a failure in runtime authorization. The Enterprise Security Gap When we give LLMs agentic capabilities (access to tools, APIs, and databases), we are fundamentally changing the security model. Traditional static permissions fail here because: 1. Context is Dynamic: An agent with read/write access might use it benignly 99% of the time, but lack the semantic understanding of when an action crosses a security boundary. 2. Lack of Intent Validation: Agents are executing multi-step API calls without runtime verification of the intent behind the payload. 3. Over-Privileged Execution: Agents often operate with static service account credentials rather than dynamically scoped, just-in-time access. Enter Single Origin: Guardrails for the Agentic Enterprise Autonomous agents require a specialized security layer. At Single Origin, we are building the infrastructure to solve this exact problem. We secure your AI pipelines by enforcing strict, programmable guardrails: 🛡️ Runtime Intent Validation: We intercept and analyze agent tool calls to ensure actions align with enterprise security policies before execution. 🛡️ Dynamic Privilege Scoping: Preventing agents from accessing or exposing sensitive data (DLP) by enforcing least-privilege access in real-time. 🛡️ Automated Failsafes: Enforcing "human-in-the-loop" approval workflows for high-risk actions. You can't scale AI autonomy without scaling your security architecture. Single Origin ensures your agents operate predictably, securely, and strictly within their designated boundaries. How is your engineering team securing agent-to-API interactions in production? Let's talk. 👇 #AIAgents #AppSec #CyberSecurity #SingleOrigin #DevSecOps #LLMSecurity

14

3 Comments

My latest CMSWire article is an update on the latest #prompt engineering techniques for #AI, featuring a few examples on Gemini and ChatGPT. "Prompts, the key instruction for large language models, are how people talk to their AI of choice. But how do people plan what to say to their AI? The starting point is #promptengineering." Read further to learn what you can do to plan better #AI usage #Tech #data #marketing https://lnkd.in/gtVnZpJr

12

3 Comments

Behold the power of Snowflake Cortex Code (#Coco). I built an entire interactive AI Workflow UX for building #Text, #Audio and #Image based AI pipelines via visual workflows using built-in Cortex AI functions. This project builds two parallel incremental pipelines. The first is a PDF Document pipeline for Legal Deposition files that extracts text, splits it to pages, translates it to Spanish, and runs sentiment scores before creating a cortex search service for a Legal Cortex agent to consume. The second is an Audio pipeline that transcribes text from MP3 files, summarizes the content, and writes it directly to tables. I am honestly not sure what is more impressive here. The new Cortex Code CLI that built this UX in a few hours with zero manual code (especially since I know nothing about React development) or the built-in Cortex AI functions that allow this interactive UX to build production grade, secure, audited and traceable AI pipelines that scale with zero effort? You tell me! Either way, consider me highly impressed by this tech & what it can do!

379

13 Comments