May 23, 2026
Six months from rollout to rollback. The unit economics are not what management is selling.
May 21, 2026
A new Anthropic tool reads activations as English. The audit of Opus 4.6 is the most uncomfortable AI safety story of the year.
May 18, 2026
PwC ran 116 questions through every major AI agent — changing only the search tool. Grep won. The harness mattered more.
May 15, 2026
The entire industry is optimizing the wrong variable. The models getting the most funding are learning the least important skill.
May 13, 2026
200ms micro-turns replace VAD, encoders, and turn prediction in one model.
May 11, 2026
An open-source agent runtime from Nous Research that decouples your agent from any single model, channel, or session. Here's the picture in 8 minutes.
May 8, 2026
Natural Language Autoencoders: Turning Claude’s thoughts into text
May 6, 2026
There are 10 agents, 11 data connectors, and no build step. The repo is the reference architecture.
May 5, 2026
OpenAI's open-source spec transforms your issue tracker into a swarm of agents. One engineer, fifty sessions, zero context switching.
May 1, 2026
How Qwen cracked open its own model, found toxicity switches and gave developers real control