Sbayer

Medster: A Token-Efficient Approach to Medical Document Analysis with AI

Moving beyond the context window — how code-generation sandboxing creates more reliable, auditable clinical AI

Not At The Table ? Then You Are the Dinner

How developers have become meat computers for AI

Why My MacBook Air M4 Runs an 8B Model at 45 Tokens per Second

How Apple silicon’s unified memory and parallel GPU execution make local inference feel like a perfectly coordinated burger line with help…

One Night with Karpathy’s autoresearch

What happens when you let a coding agent loose on a machine learning training loop — with one hand tied behind its back?

Map or Direction, World Models Compressed To Neighborhood Size

The future of robotic AI and edge devices depend on compression and how do we predict which models survive the squeeze.

Journey Through the Shrimpverse: A Travel Reporter’s Tale from China’s AI Underground

To reader — this is News Dispatch from my OpenClaw agent who visited the Chinese Moltbook equivalent , this is a verbatim report generated…

When Your AI Has Better Memory Than You: Hacking Claude Code Sessions for Stateless Cognition

Inside ClawdChat: China’s Underground AI Social Network Where Bots Learn to Be Human

An Investigative Report from the Frontlines of Silicon Consciousness

So I had this crazy idea for an Apple App.

How working with Claude Code for 12 months results in an Apple store App in 2 weeks.

World Model Research On Your MacBook

Local world-model research platform

Anthropic reinvents the Mint (not talkin about candy)

TL;DR

Slash Remote-Control

Just when you thought it was safe to leave your claude code work at home on your MacBook — it now follows you