Alphabell.

AI makes AI better.

A daily hub for work where AI is part of the loop that improves AI, each entry with its loop spelled out. Read more →

1/4
A loop in motion
AlphaEvolve
Project · Google DeepMind
Gemini models propose code changes
From the labalphaBell Qwen Open27B post-training research, with capability tradeoffs measured.Explore the model → From the labModel ObservatoryMeasured Flint and Qwen traces with published decoder maps.Open the lab tool → From the labMiniMax Video BudgetVisualize sampled frames, two-frame groups and vision placeholders.Try the inspector → From the labMOSS Blind Layout CheckCompare exact text and placement in two local images.Try the blind check → From the labAgent Request Evidence LadderSee what a trace proves about an agent's request and its outcome.Explore the cases →

Today's radar

19 Nov 2026Full radar →

7 picked from 136 candidates · 30 sources read · 79 loops indexed

  1. 01
    Loop of the dayArticleHacker News18 Nov 2026Self-modifying agents

    Show HN: Mastra Factory, the open-source software factory we use to build Mastra

    The loop: Mastra's durable agents operate a software factory that opens and merges pull requests across the project's codebase. This work directly maintains and improves the Mastra framework, which is the very tool used to build and run the factory.

    Mastra Factory demonstrates a practical, deployed loop where an open-source agent framework is actively used to merge hundreds of pull requests a week to maintain and improve itself.

    loop fit 9/10codekaratevia Hacker News
    Show HN: Mastra Factory, the open-source software factory we use to build Mastra
  2. 02
    PaperarXiv19 Nov 2026Self-reward and self-play

    SWE-growth: Teacher-Guided Adaptive Self-Play for Software Engineering Agents

    The loop: A student coding model attempts repository-level repair tasks, and a teacher model identifies learning opportunities from the failures. These verified outcomes are used as training data for supervised fine-tuning and reinforcement learning, making the student model better at software engineering.

    loop fit 10/10Songshi Liang, Zixuan Huang, Mingxuan Song, Haoyu Tang, Shusheng Xu, Tongkai...via arXiv
    SWE-growth: Teacher-Guided Adaptive Self-Play for Software Engineering Agents
  3. 03
    Blog postOpenAI18 Nov 2026Hardware and chips

    Advancing chip design with OpenAI models

    The loop: OpenAI models optimize circuits and RISC-V core designs for silicon. These hardware improvements are then used in the chips that run the models, accelerating their future execution.

    loop fit 9/10via OpenAI
    Advancing chip design with OpenAI models
  4. 04
    ProjectGitHub2 Nov 2026Tools and harness

    lisresearchgroup/math-rsi

    The loop: A self-improving agent uses a research loop and verifiable proofs to make mathematical discoveries. These verified proofs expand the agent's memory and proof capabilities, making the harness more effective for further mathematical discovery.

    loop fit 8/10lisresearchgroupvia GitHub
    lisresearchgroup/math-rsi
  5. 05
    PaperarXiv19 Nov 2026AI R&D evals

    RSIBench-Infra: Can Agents Optimize Their Own Training and Inference Infrastructure at Production Scale?

    The loop: Autonomous agents modify the software stack across engine, kernel, and instruction levels to optimize training and inference workloads. This improves the efficiency of the infrastructure, allowing the same agents to run or train faster in subsequent iterations.

    loop fit 9/10Hanlin Wu, Yifei Wang, Zhicheng Cai, Huan-ang Gao, Wei-Ying Ma, Ya-Qin Zhang,...via arXiv
    RSIBench-Infra: Can Agents Optimize Their Own Training and Inference Infrastructure at Production Scale?
  6. 06
    PaperarXiv19 Nov 2026Inference efficiency

    QDA: Closed-Loop LLM Quantization Design with Agents

    The loop: An LLM agent designs and evaluates post-training quantization recipes for large language models. This improves the inference efficiency of those models, making the agent's own execution or its successor's deployment faster and more cost-effective.

    loop fit 9/10Xingyang Li, Tianlang Zhao, Song Han, Ligeng Zhuvia arXiv
    QDA: Closed-Loop LLM Quantization Design with Agents
  7. 07
    PaperarXiv19 Nov 2026Kernels and compilers

    Arbor: How Search Policy Shapes Agent Behavior in System Optimization

    The loop: Autonomous agents search for and write kernel-level and framework-level optimizations across an explicit optimization tree. These optimizations speed up the training and inference of AI models, improving the execution of the agents themselves.

    loop fit 9/10Neha Prakriya, Chaojun Hou, Zheng Gong, Huasha Zhao, Yanyuan Qin, Chris Cai,...via arXiv
    Arbor: How Search Policy Shapes Agent Behavior in System Optimization
  8. 08
    PaperarXiv18 Nov 2026Self-modifying agents

    Language Models Can Rewrite Their Own Weights

    The loop: A 27B language model diagnoses its own failures and directly edits its own weights without gradients or training. These edited weights are then used by the model to achieve higher accuracy on subsequent unseen problems.

    loop fit 10/10Jiajun Fan, Neil He, Changyi Yang, Andy Mao, Xi Zhu, Chaoran Cheng, Kaiqu...via arXiv
    Language Models Can Rewrite Their Own Weights
AlphaEvolve
ProjectGoogle DeepMind
AlphaEvolve
  1. Gemini models propose code changes
  2. Evaluators keep best scoring versions
  3. Finds better Gemini training kernels
  4. Lowers cost to train next Gemini models
  5. repeat
Kernels and compilersGoogle DeepMind · 2025
Darwin Gödel Machine
PaperarXiv
Darwin Gödel Machine
  1. Coding agent edits its own code
  2. Scored child agents enter the archive
  3. Archived agents make later self modifications
  4. repeat
Self-modifying agentsUBC, Vector Institute, Sakana AI · 2025
AlphaChip
ProjectGoogle DeepMind
AlphaChip
  1. RL agent lays out TPU blocks
  2. Better layouts make AI hardware cheaper
  3. Agent pre-trains on earlier chip generations
  4. repeat
Hardware and chipsGoogle DeepMind, Google Research · 2020
autoresearch
Mini-projectGitHub
autoresearch
  1. Coding agent edits LLM training script
  2. Runs a 5-minute training job
  3. Keeps change if validation improves
  4. Changes transferred to larger models
  5. repeat
Research automationAndrej Karpathy · 2026
ScholarEvolve
PaperarXiv
ScholarEvolve
  1. Research agent writes harness mutation blueprints
  2. Coding agent implements harness module changes
  3. Winning harness becomes the new champion
  4. New champion harness runs further generations
  5. repeat
Tools and harnessYang et al. (UC Santa Barbara, Microsoft) · 2026
Nemotron-4 340B
PaperarXiv
Nemotron-4 340B
  1. Aligned model generates training data
  2. Reward model judges and filters it
  3. Data trains next intermediate instruct model
  4. Next model generates better alignment data
  5. repeat
Training dataNVIDIA · 2024
Self-Rewarding LMs
PaperarXiv
Self-Rewarding LMs
  1. Model generates candidate responses to prompts
  2. Model scores them itself as a judge
  3. Scores become preference pairs for DPO
  4. Trained model becomes next generator and judge
  5. repeat
Self-reward and self-playMeta, NYU (Yuan et al.) · 2024
Claude Code
ProjectAnthropic
Claude Code
  1. Claude Code writes its own code changes
  2. Changes become next Claude Code versions
  3. Release becomes harness for next development round
  4. repeat
Self-modifying agentsAnthropic · 2025
KernelEvolve
ProjectarXiv
KernelEvolve
  1. LLM agents write and benchmark kernels
  2. Kernels speed up AI model training
  3. Session data post-trains smaller kernel models
  4. repeat
Kernels and compilersMeta · 2025
PostTrainBench
BenchmarkarXiv
PostTrainBench
  1. Coding agent writes fine-tuning code
  2. Agent post-trains a base LLM
  3. Trained model scored on target benchmark
  4. One turn of model training another
  5. repeat
AI R&D evalsELLIS Institute Tubingen, MPI-IS, University of Tubingen, Thoughtful Lab (Rank et al.) · 2026
Self-training learned optimizers
FoundationarXiv
Self-training learned optimizers
  1. Learned optimizers update other learned optimizers
  2. Training keeps best performing optimizer parameters
  3. Better optimizers train each other faster
  4. repeat
Architecture and optimizer searchMetz et al. (Google Research, Brain Team) · 2021
CoT monitoring
PaperarXiv
CoT monitoring
  1. LLM reads reasoning model thoughts
  2. LLM flags reward hacking behavior
  3. Signal feeds back into training reward
  4. Trains more aligned capable agents
  5. repeat
Interpretability and oversightOpenAI (Baker et al.) · 2025

Support for loop research

How to apply →

Have a loop worth building?

Tell us what the AI improves and how the improvement comes back. Applications for grants, fellowships and compute all start on the contact page.

Get in touch