Thoughts on AI, engineering, and building
Practical writing on AI agents, automation, and the engineering behind them, for the people who run the business.
Security & Governancebrief
Five practical permission rules for AI agents, drawn from joint cybersecurity guidance on agentic AI and analysis of two agent-framework vulnerabilities.
· 2 min read
Agentsexplainer
A plain-English primer on physical AI: how it differs from conventional industrial automation, what the new robot models do, and what to ask before piloting.
· 3 min read
Models & Benchmarksnews-analysis
OpenAI, Anthropic and Moonshot each shipped a major model in July. What matters for buyers is the tiering, the effort controls and the cost per task.
· 4 min read
Enterprise Adoptiondeep-dive
Deloitte's executive survey and IFR robot counts give a modest view of AI in factories: strong intent, early deployment, and a human-heavy floor.
· 3 min read
Security & Governancechecklist
A short checklist for adding AI failures, prompt injection and model withdrawal to the incident-response plan you already have.
· 2 min read
Hardware & Infrastructurehardware-breakdown
AI systems now ship as liquid-cooled racks of 72 GPUs on one fabric. Published GB300 NVL72 and Vera Rubin NVL72 specs, and what they ask of a data hall.
· 4 min read
Security & Governanceexplainer
Guardrails screen what goes into and out of an AI system. A plain-English guide to the five checkpoints, what the evidence says, and where they fall short.
· 3 min read
Models & Benchmarksbrief
ByteDance says Seedance 2.5 makes 30-second clips with joint audio; Google's Veo 3.1 page cites 8-second clips. All claims are vendor-stated.
· 2 min read
Enterprise Adoptionnews-analysis
Two June 2026 surveys put roughly half of organizations at agents in production. The more useful numbers are about cost control, tracing and evaluation.
· 3 min read
Enterprise Adoptionnews-analysis
Six months of sourced figures: spending, adoption, agent benchmarks, model pricing and the risks that came with them.
· 3 min read
Harnesses & Toolingexplainer
Test the retrieval step of a RAG system on its own: recall@k, precision@k, MRR and nDCG, a small worked example, and how to build a golden question set.
· 4 min read
Enterprise Adoptionbrief
Gallup's survey of 23,717 U.S. employees links frequent AI use to manager support, clear policies and tools that fit existing systems.
· 2 min read