AutoResearch has been public for weeks and quickly became one of the fastest-growing repositories on GitHub. Community forks support the RTX , Apple Silicon via MLX, and multi-GPU setups via SkyPilot. Tobi Lutke demonstrated that the loop works on non-ML codebases.
You'll notice the tools are converging. AI Scientist-v writes papers. AIDE wins Kaggle competitions. Robin runs biology experiments. DSPy optimizes prompts. All use the same core loop you've learned: propose, test, measure, decide.
The bottleneck has shifted from execution to hypothesis quality. Running experiments is cheap. Knowing which experiments to run is not. Your role has moved from writing code to writing research instructions.