Long-form articles and engineering guides, newest first. Older reference material also lives in the technical archive.
DevOpsWhat changed
Three small changes landed in the last two weeks. None is a crisis. Each can quietly break a pipeline or leave a service on a runtime that no longer gets fixes.
Artificial IntelligenceWhat changed
Sakana AI's MASS research explores how one model can learn from the work of multiple agents. We examine the process, results and limits.
Artificial IntelligenceFoundation
CPUs, GPUs, TPUs, NPUs, LPUs and DPUs do different kinds of work. Understand each processor in simple language, with everyday examples and a comparison table.
Artificial IntelligenceWhat changed
HCLTech and Anthropic plan to integrate Claude into enterprise services and software. Here is what each company hopes to gain and what remains unproven.
Career & LearningWhat changed
Millions of users and growing revenue did not prevent Character.AI's bankruptcy. A technical and business look at inference costs, monetization, and operational responsibilities.
Career & LearningWhat changed
Research on AI coding paints a mixed picture. A payment-retry example shows why passing tests and generating code quickly are not the same as delivering reliable software.
CybersecurityWhat changed
An investigation into how connected refrigeration works, what researchers found in Danfoss controllers and why cybersecurity can damage physical operations.
CybersecurityWhat changed
Researchers traced PoeLLM to vulnerable servers running AI and developer tools. Understand the exploit path, unusual GitHub-poem technique and security lessons.
Artificial IntelligenceFoundation
Huawei, Alibaba, Baidu and Tencent are building different parts of China's AI computing ecosystem. Here's how the technology works and what remains unproven.
Artificial IntelligenceWhat changed
Microsoft plans to scale Maia infrastructure in 2027. Learn what Maia 300 changes in memory, networking and power, and what its headline numbers do not prove.
Software & SystemsWhat changed
A Copilot plugin may look like one feature while relying on separately operated APIs, identities and services. Who maintains those dependencies when something fails?
DevOpsWhat changed
AI incident-response agents can investigate failures, but production permissions and engineering judgement still shape safe remediation.
DevOpsGuide
Passing validation and plan checks does not prove infrastructure meets security requirements. Here is how to choose the right native tests and independent checks.
Cloud ComputingFoundation
SeriesPart 2 of 2Closing the laptop should not lose the assignment. What must enterprise infrastructure preserve when an AI agent waits, fails, or reaches its budget?
Software & SystemsFoundation
SeriesPart 1 of 2AI agents can investigate problems and determine their next actions. But enterprise systems still need to control permissions, transactions, and execution.
Software & SystemsFoundation
AI introduces new requirements for testing, monitoring, and model management. Learn how these responsibilities extend the traditional software development lifecycle.
Artificial IntelligenceWhat changed
Microsoft is bringing local AI, cloud models, agents and new security controls together in Windows. Here is what hybrid intelligence could mean for personal computing.
Artificial IntelligenceFoundation
SeriesPart 2 of 2Follow one customer delivery problem through a production AI agent: context, model decisions, tools, state, retrieval, authorization, failed actions and evidence. See why the model is one component inside a larger engineered system.
Artificial IntelligenceFoundation
SeriesPart 1 of 2What really separates Generative AI, AI agents, and Agentic AI? By following a single customer delivery problem from content generation to autonomous action, see how AI changes as it gains goals, tools, and decision authority.
Software & SystemsFoundation
Coding agents can understand a repository and still make the wrong decision. Important engineering knowledge often lives in policies, documentation, operational systems and people's experience, not in the code itself.
