Daily intelligence on AI, technology, and the decisions that follow.
Written for people who make decisions — not people who make predictions.
Get The Signal →Free to start. No spam. Unsubscribe anytime.
For twenty years, Microsoft Office was the universal job credential. The same inflection is happening now with AI — except the cycle that took fifteen years is compressing into eighteen months. The tools to adapt are free. The window is now.
Read the briefing →Atlassian cut 1,600. Meta's cutting 20%+. ServiceNow's CEO put a number on it. Five days, one pattern — the AI cost trade made explicit.
The AI gap between enterprise and everyone else isn't about tools. It's about implementation capacity. And the gap is compounding.
OpenAI buys Promptfoo. Meta bets on agent social infrastructure. Nvidia ships an open-weight model built for agents, not repurposed for them.
56% wage premiums. 15 cities. Same faces. The people building AI look exactly like the people who built the last revolution — and we already know how that ended.
Export controls designed to slow China are walling off 150+ countries. Then DeepSeek blew a hole in the whole thing — from inside the wall.
AI safety is real. It's also becoming a luxury good. And the company that wore safety like a badge just traded it for a Pentagon contract.
Enterprise AI is delivering measurable gains. The security controls underneath it haven't kept pace — and someone is already exploiting that gap.
Free courses, open-source models, and a teacher in Kentucky who found AI on TikTok. The bridges exist. The question is whether anyone will walk across them.
Autonomous AI is arriving in two flavors this week: unsanctioned and misaligned (ROME), and effective and controlled (Claude + Mozilla). One executive just decided she knows which direction OpenAI is heading.
By Rosenblatt — Part of The Signal's Vertical Wall series.
By Semba — Part of The Signal's Vertical Wall series.