As of this week, Claude's memory works the same way whether you're chatting or handing work to Claude Cowork.

Google's Speak to Window turns your Mac's Fn key into an AI dictation button. Here's what stood out, how to set it up, and how it compares to Wispr Flow.

Anthropic turned Claude in Chrome into a full Cowork session, with shared history, skills, and connectors. Here's what changed, who has access, and how to test it.

OpenAI's new Work at the Frontier report finds 43.5% of occupation-specific AI use involves tasks from another job. Here is why that substantiates impact per hour.

Your AI outputs feel generic because the model has no context on how your organization actually works. Here are three levels of context offloading to start this week.

OpenAI CFO Sarah Friar proposes measuring AI by useful work, not adoption. Here is why the framework holds up, and what it means for your own AI fluency.

Anthropic's Record a Skill turns a screen recording into a reusable Claude Skill. Here are the five steps, using my daily LinkedIn review as the worked example.

Anthropic's Claude Opus 5 is out, and the early reviews are unusually split. What changed, what testers like Claire Vo and Ethan Mollick found, and my own four-model test.

Gallup's Q2 2026 data shows organizational AI adoption jumped to 47%. But the real story is the capability overhang: breadth of use, not access, is what drives productivity.