Karpathy open-sources AutoResearch: AI agents that run the full research loop autonomously
mdancho84 · x · 2026-10-22
Andrej Karpathy has open-sourced AutoResearch, a system of AI agents that autonomously run the research loop — not just answering questions but proposing, executing and iterating on experiments.
A progress chart shows 83 experiments with 15 kept improvements, driving validation Brier scores from 0.998 down to 0.977, with auto-generated rationales for each accepted change (model swaps, sampling-frequency tweaks, prompt edits).
The README opens with a tongue-in-cheek note from 'Karpathy, March 2026': frontier research once done by 'meat computers' between group meetings is now the domain of autonomous agent swarms on compute clusters, with self-modifying code beyond human comprehension.
More from coding & agent
- mdancho84: The Valuable Ones Architect Systems — Agents, Workflows, Memory, Eval Loops — mdancho84 · 2026-10-22
- AutoResearch: AI Agents That Run the Full Research Loop Autonomously — mdancho84 · 2026-10-22
- Dev uses Claude Opus 5.5 to build a working replica of ILM's Stagecraft volume in Three.js — jasonkneen · 2026-10-22
- Solo founder ships one-command tool to find X's biggest Claude Code and Codex token burners — ThePeterMick · 2026-10-22
- Open-Source Claude Recruiter Finds Ranked Candidates with Personal Emails at $0.20 Each — brandon_galang · 2026-10-22
- Retention, not loops: why recursive self-improvement keeps starting over — BandAutomatic2009 · 2026-10-22