The librarian in the loop
A developer gave Claude Code access to 100 books and a simple command: "find something interesting." What came back wasn't summaries. It was connections no hand-tuned pipeline could find.
Samsung pledges to double AI devices to 800 million while warning chip shortages will force price hikes. The contradiction reveals how memory costs are killing hardware margins—pushing the company toward insurance data monetization instead.
Meta's chief AI scientist just gave an interview that undercuts the tidy narrative. What it reveals about benchmark gaming, leadership panic, and the dead end of language model scaling.
A German indie developer spent 18 months building AI autocomplete that works everywhere on your Mac. It runs locally, costs nothing in cloud fees, and faces one existential threat: Apple announcing the same feature as a macOS bullet point.
Instagram's Adam Mosseri admits Meta can't detect AI content flooding the platform—and says camera manufacturers should solve the problem Meta helped create. Photographers face a choice: degrade their work to prove they're human, or get buried by free synthetic content.
Anthropic's multi-agent Claude Code lets developers run five AI assistants simultaneously. Early adopters report 10x productivity gains. The token bills tell a different story about who really benefits from autonomous overnight coding.
Meta spent billions building free AI that nobody pays for. Then it bought Manus, a startup that convinced millions to subscribe in eight months. The acquisition reveals more about Meta's monetization struggle than its technology.
AI coding tools promise 55% productivity gains. A new study found developers actually work 19% slower with them—but feel faster. The gap between perception and reality explains why some users can't stop prompting, even at 2 AM.
Astra AI just won Slovenia's Startup of the Year. The numbers look great: 170,000 users, expansion into Germany, rave reviews. But their claimed 50 billion monthly tokens raises a question no one's asking: can a €24/month subscription cover those API bills?
Apple buried RDMA over Thunderbolt in a beta update. Now four Mac Studios can run trillion-parameter AI models at conversational speed. The cost: $50,000. Equivalent NVIDIA hardware: $780,000. No press release. No keynote. Just a checkbox in recovery mode.
China's EUV prototype isn't a technological defeat for the West. It's a counterintelligence one. The vector isn't smuggled crates. It's people. Europe discovered, again, that openness without defense is vulnerability, not virtue.
Chinese scientists built a working EUV prototype using former ASML engineers and secondary-market parts. The machine generates light but hasn't produced chips. ASML took 18 years from prototype to production. Beijing wants 3-5. The math doesn't add up.
Google calls Gemini 3 Flash "a fraction of the cost." Compared to Pro models, sure. Compared to the Flash model it replaces? Prices rose 67%. The real story: what happens when the industry stops subsidizing API access.
Get the 5-minute Silicon Valley AI briefing, every weekday morning — free.