Everything we've published, newest first. Covers AI, software development, security, hardware, and cloud.
We spent a month using four of the top AI coding assistants on production work. Here's what each one is actually good at, where they break, and which to pick for your team.
Local models finally caught up to something useful. Here's a real-world walkthrough of getting Llama 3, Qwen, and others running on consumer hardware — and when it's worth doing versus staying in the cloud.
We ran the same 50 queries through each of them. The results are messier — and more interesting — than the marketing suggests.
Reasoning models changed the rules. A lot of the old advice ('be very specific,' 'use few-shot examples') is now outdated or actively harmful. Here's what we've learned.
Strip the hype away and 'agents' are LLMs in a loop with tool access. That's powerful — but only in specific shapes of problem. Here's our real-world take.
A practical walk-through of replacing a real, boring task — inbox triage — with an LLM-driven script. Not theoretical. Just the working code and what tripped us up.
Monorepo tooling has quietly gotten much better. We set up the same project three ways to compare cold-install time, CI speed, and developer ergonomics in 2026.
'Zero trust' gets thrown around as a buzzword. For a 5–20 person engineering team, here's what it actually looks like — and which three things to tackle first.
The laptop wars got interesting again in 2026. We compare the three most credible choices for engineers — including thermals, repairability, and the weird gotcha specs manufacturers hide.
Hosting AI-powered apps is a different beast: streaming responses, long-running requests, unpredictable cost. Here's how each platform handles it, with real numbers.
Six months of running Claude-powered agents across our ad-tech stack. The wins were smaller than the hype promised. The failures were more boring than the horror stories. Here's what actually moved the needle.
Opinions from our team on the Claude Code features that actually change how we work: sessions, skills, hooks, subagents, and the CLI flags that made us faster.