AI · 15 min read
Everything known about Gemini 4 Argon: Google's benchmark claims against GPT-6 Astra and Claude, what independent tests show, real cost per task, the 1M-token output limit, and who can use it.
Software · 12 min read
What decides whether ChatGPT search, Google AI Overviews, AI Mode, Copilot and Claude can cite your site: the crawlers, robots.txt, CDN bot rules, page structure and how to measure it.
AI · 15 min read
Claude Opus 5.5 against GPT-6 Astra and GPT-6 Sol: where the vendors' own benchmark tables disagree, what independent tests show, cost per task versus price per token, and which to pick.
Security · 8 min read
ShinyHunters says it breached the FBI through an Oracle PeopleSoft zero-day. What the FBI has confirmed, what remains a claim, and the steps PeopleSoft operators should take now.
AI · 7 min read
How the Model Context Protocol works: hosts, clients and servers, the JSON-RPC data layer, tools, resources and prompts, transports, the 2026-07-28 stateless redesign, and the security model.
Cloud · 5 min read
DORA's five software delivery metrics, what each one measures, where the data comes from, and how to use them to improve delivery without turning them into targets that teams learn to game.
Software · 14 min read
A method for choosing a software development partner: define the problem, pick the contract model, verify vendors in public registers, and protect ownership and data. With worked examples from Norway.