IBM shares plunged 13% after Anthropic revealed that Claude Code can accelerate COBOL modernisation—the kind of expensive legacy work that underpinned IBM's consulting revenue for decades. AI is finally making legacy system upgrades solvable.
AI models are amazing at many tasks, but parsing complex PDFs remains a stubborn problem—especially when thousands of pages need to be sifted through quickly.
HBO's medical drama explores the tension between AI-powered transcription tools and human oversight in emergency rooms, showing how technology can't fix systemic understaffing issues.
By the end of 2025, Instagram’s Adam Mosseri warned that authenticity is being eroded by AI, sparking a debate about how seriously Big Tech is addressing the problem. This post breaks down the stakes, responses and what it means for creators and users alike.
A new open-source AI coding assistant called Goose offers similar capabilities to Anthropic's Claude Code—but without the hefty $200/month price tag.
Anthropic has accused DeepSeek and two other Chinese AI startups of running massive, illicit data‑driven campaigns that secretly leveraged its Claude model to train rival systems.
Boris Cherny’s new workflow shows a single engineer running five Claude agents in parallel can outpace a whole dev team. The secret? A slow, smart model, shared knowledge files, and automation that lets AI test, refactor, and ship code on its own.
Nous Research’s new NousCoder‑14B shows that a 14‑billion‑parameter model trained on 24,000 verified problems can rival proprietary giants in just four days of compute – all with a fully open stack that anyone can replicate.
Anthropic’s new Cowork lets anyone—not just developers—tell Claude to read, edit, and create files inside a sandboxed folder, delivering a hands‑off productivity boost that feels like a real coworker.
