New post

Two companies, one market, completely different strategies.

Claude Opus 4.8 Is Here — And Honesty Is Now a Benchmark

SWE-bench Pro hit 69.2%. But the 0% uncritical fault rate is the real story.

Google's secret weapon just became a product

Gemini Spark finally turns Gmail into an AI agent advantage — here's what I actually think about it.

OpenAI's Codex update: Goal mode is now GA

A desk review of what shipped on May 21-22 - and what I'd actually do with it.

Trying Gemini 3.5 Flash from Google I/O 2026 — the parts you can use for free

Claude Opus 4.7: A Quiet Upgrade That Earns Its Keep at Work

I ran four ordinary business tasks through it. The difference was clearer than the version number suggests.