Clauday.

Yesterday the field finished shifting its centre of gravity: the thing being optimised is no longer the model, it is the harness the model runs inside. The same GPT-5.6 Sol went fr…

Two things collided today. DeepSeek shipped a harness where the agent loop itself is a swappable plugin — and where an agent can inspect its own runtime, write a new plugin, mount …

The most useful thing posted today was a receipt. Someone spent real money benchmarking the idea everyone repeats casually, put Claude Code in a loop and let it improve your agent,…

Four of the most senior technical people at the company that invented the transformer walked out of Google to build an autoresearch startup, with Alphabet as a founding investor. T…

Today the loop stopped being a technique and started being an infrastructure problem. Microsoft published the first production-scale measurement of what agentic coding actually doe…

Prime Intellect open-sourcing Prime Agent owns the day: a dozen separate posts dissect its recursive-language-model design, the persistent IPython kernel that turns context into va…

Prime Intellect's Prime Agent landed hard enough to reorganize the whole conversation: a self-improving RLM harness where context is a variable the model programs over, sub-agents …

The autoresearch conversation on August 6 finally started producing numbers instead of vibes: shaved unit test runtimes, halved agent step counts, kernel latencies driven into loca…

The loop world got its biggest product launch in weeks: Prime Intellect shipped Prime Agent, a self-improving harness whose headline number — Opus 5 jumping from 30.2% to 95.5% on …

The loop conversation split into two clear camps today. On one side, working loops shipping real artifacts: a GPS-free drone navigation model invented from scratch in 81 iterations…