Huge week for autonomous coding agents:
Google released a self-coding agent.
& it ended up solving several maths problems that humans couldn’t.
it also improved the performance of the chips + model it runs on.
all autonomous - no human interference.
OpenAI launched Codex
which now lets you code multiple tasks simultaneously
just link your codebase, tell it what to do & continue on with your life.
Engineers have already cut down weeks of work to days.
OpenAI released 4.1
Their new model is better at coding than 4o & faster than o3 (though not at good)
Cannot believe the rate at which these agents are improving.
We’re reaching a point where these things are not just going to be autonomous
but smarter at doing the work us humans do.
So cool