AI — Page 23

Claude Code Now Supports Nested Subagents

Anthropic added nested subagent support to Claude Code. If you're already using subagents in your workflows, here's what changes—and what new risks come with it.

Large "LIFE OS" text with arrow pointing to four blue-outlined boxes listing Memory, Skills, Workflows, and Goals against a…

PAI Gives Claude Code Persistent Memory and Structure

AI. Yuki Okonkwo3 months ago· 7 min read

PAI Gives Claude Code Persistent Memory and Structure

PAI adds persistent memory, custom skills, and structured workflows to Claude Code. Here's what it does well, what it costs you, and who actually needs it.

Man in blue shirt discusses AI model optimization with overlaid graphics showing rLLM-FinQA-4B training framework and Qwen3…

A 4B Model Beat a 235B Model for Under $500

Samira Barnes·3 months ago

A 4B Model Beat a 235B Model for Under $500

Snorkel's Kobie Crawford shows how a 4B parameter model outperformed Qwen 3 235B on financial analysis tasks using RL training that cost less than $500.

Who Owns What AI Says About Your Brand?

AI. Samira Barnes3 months ago· 7 min read

Who Owns What AI Says About Your Brand?

AI tools describe brands to consumers millions of times daily. No regulator has decided who's accountable when those descriptions are wrong. That gap is the real story.

A futuristic armored figure stands in a dark cavern with glowing green accents and text overlay reading "NEURAL DAWN" and…

Arm's Next Mali GPU Will Carry Neural Accelerators

Bob Reynolds·3 months ago

AI. Bob Reynolds3 months ago· 6 min read

Arm's Next Mali GPU Will Carry Neural Accelerators

Arm's Neural Dawn demo signals dedicated neural accelerators in the next Mali GPU generation, bringing desktop-style upscaling and dynamic lighting to Android.

Building a Serverless AI Agent with Pi and Google Cloud

Building a Serverless AI Agent with Pi and Google Cloud

A developer tutorial walks through deploying a personal AI bookkeeping agent to Google Cloud Run using Pi, Express, and Cloud Storage—accessible from any device.

Cream-colored background with "Fable 5 by Anthropic" text on left and decorative vintage butterfly illustrations arranged…

Claude Fable 5 Launches With Tight Safety Guardrails

Nadia Marchetti·3 months ago

Claude Fable 5 Launches With Tight Safety Guardrails

Anthropic's Claude Fable 5 is out, but safety restrictions, a data retention shift, and subscription changes make the launch more complicated than the benchmarks suggest.

A woman in a maroon shirt appears skeptical next to a gold-and-blue quantum computer chip labeled with Microsoft branding…

Microsoft Majorana 2: Bold Claims, Thin Evidence

AI. Nadia Marchetti3 months ago· 6 min read

Microsoft Majorana 2: Bold Claims, Thin Evidence

Microsoft's Majorana 2 chip promises a 1,000x reliability leap and a 2029 quantum computer. Scientists say the paper doesn't back any of it up.

Man with glasses beside a folder icon showing bidirectional arrows between a red sparkle app and blue coding app labeled…

AI Coding Loops Are Replacing the Prompt—Now What?

Samira Barnes·3 months ago

AI. Samira Barnes3 months ago· 7 min read

AI Coding Loops Are Replacing the Prompt—Now What?

Developers are designing autonomous AI loops that merge code without human review. The engineering logic is sound. The accountability framework is nonexistent.

OpenAI Plans to Remake ChatGPT as an AI Super App

OpenAI Plans to Remake ChatGPT as an AI Super App

OpenAI is reportedly overhauling ChatGPT into a full AI super app with coding agents, automation, and new security features. Here's what's verified and what's still reported.

Two smiling engineers wearing conference badges flank the Cloudflare and AI Engineer Europe logos against a warm gradient…

Cloudflare's Dynamic Workers Rehabilitate eval()

AI. Yuki Okonkwo3 months ago· 8 min read

Cloudflare's Dynamic Workers Rehabilitate eval()

Cloudflare's Sunil Pai and Matt Carrie explain how Durable Objects and Dynamic Workers form a new compute foundation for AI agents—and why eval() deserves a second look.

Man in checkered blazer standing before technical architecture diagrams showing DeepSpeed Ulysses and Untied Ulysses…

AI Agents With 5M-Token Memory Raise Privacy Questions

AI Agents With 5M-Token Memory Raise Privacy Questions

As AI agents gain the ability to hold millions of tokens in context, Rach Kovacs examines what that means for user privacy, data retention, and security exposure.

Four men's headshots displayed against black background with yellow header text reading "AI Personhood Gets Granted" and…

Anthropic Wants a Pause. Argentina Wants AI Companies.

AI. Yuki Okonkwo3 months ago· 7 min read

Anthropic Wants a Pause. Argentina Wants AI Companies.

Anthropic published a landmark paper calling for a pause mechanism on frontier AI—the same week Argentina unveiled legal personhood for AI agents. Two signals, one reckoning.

Ten Claude Code Plugins Worth Adding in 2026

Ten Claude Code Plugins Worth Adding in 2026

From knowledge graphs to adversarial code review, Chase AI maps ten Claude Code plugins that address real workflow gaps—not just hype.

AI Agents Now Build and Fix Their Own Web Scrapers

AI Agents Now Build and Fix Their Own Web Scrapers

AI agents can now build, run, and repair web scrapers without human input. Here's what that pipeline looks like—and what it means for everyone online.

Neon "NEW & FREE" text surrounds a glowing rainbow star icon on a circuit board with electric lightning effects in vibrant…

Google's Six New AI Tools: What They Do and Who They're For

AI. Yuki Okonkwo3 months ago· 7 min read

Google's Six New AI Tools: What They Do and Who They're For

Google shipped six AI tools at once—Imagen 3, Gemma 4 12B, Magenta Realtime 2, Co-scientist, Dream Beans, and quantized Gemma 4. Here's what each actually does.

Man in maroon sweater gesturing while speaking in a home office library setting with "TEST FIRST" text overlay in white and…

Test-Driven Development Tames AI Coding Agents

Alex Volkov·3 months ago

AI. Alex Volkov3 months ago· 8 min read

Test-Driven Development Tames AI Coding Agents

Red-green-refactor is a decades-old discipline. Brainqub3 argues it's also the most practical fix for AI coding agents that break things while fixing things.

A tablet with keyboard displaying "27" on its screen against a vibrant blue and yellow gradient background with a clock…

iPadOS 27's AI Features Previewed Before WWDC

AI. Rachel "Rach" Kovacs3 months ago· 6 min read

iPadOS 27's AI Features Previewed Before WWDC

Four rumored iPadOS 27 features—AI Safari tabs, a revamped Spotlight, natural language shortcuts, and a grammar tool—suggest Apple's next iPad update prioritizes intelligence over spectacle.

AI's Economic Impact: Jobs, Tasks, and the Iceberg

Samira Barnes·3 months ago

AI. Samira Barnes3 months ago· 8 min read

AI's Economic Impact: Jobs, Tasks, and the Iceberg

A new MIT index reveals AI's economic exposure is five times larger than headlines suggest—and concentrated in places no one is watching. Here's what the data shows.

Neon orange padlock with glowing burst symbol chained shut against dark background, with "Leaked." text and arrow pointing…

Claude Mythos: Hype, Leaks, and What Anthropic Said