The Deep End

09.24

[2609.24972] RRSI: Regularized Recursive Self-Improvement of Agent Harnesses (arxiv.org)

09.23

tokens too cheap to meter (jyn.dev)

Claude discovers a novel enzyme system \ Anthropic (www.anthropic.com)

Introducing Strands harness: frontier performance with 28% lower token cost... (strandsagents.com)

Meet Stripe's Knowledge AI Platform | Stripe Dot Dev Blog (stripe.dev)

09.22

Open-Weight Inference Economics & GPU Useful Life | Ornn Data (data.ornn.com)

Step 2. Token Embedding (poloclub.github.io)

09.20

Trying the Software factory pattern. | Irrational Exuberance (lethain.com)

阶跃星辰 (www.stepfun.com)

the senior engineer death spiral • Solving the decision problem (sunilpai.dev)

Pirate Face - Turn AI into torrents that live forever (pirateface.co)

Poster Prompts | ‘ERE I AM - JH! (ukslim.github.io)

09.19

I Built Non-Autoregressive Decision Models with RL a Year Ago | Hacker News (news.ycombinator.com)

Laya — 33ms Multilingual System 1 Decision Engine with Calibrated Probabilities (laya.convaiinnovations.com)

09.18

AI chatbots are becoming experts at changing people's minds. What's their... (www.science.org)

Needle 2 - The 14 MB Agentic LLM for Tiny Devices | Cactus (cactuscompute.com)

[2609.20804] An Empirical Study of Harness Design for Coding Agents (arxiv.org)

09.17

The Anthropic Glasswing Receipts Are Starting to Trickle In | Blog | VulnCheck (www.vulncheck.com)

Meat proxy - sf-isms (sfisms.org)

Towards Self-Driving Codebases | Detail (blog.detail.dev)

Economic Policy for AGI — DeepMind Institute (institute.deepmind.com)

Toward Recursive Self-Improvement: How GLM Built Its Own Inference Infrastructure (z.ai)

[2609.18842] Infinite-Parameter LLMs: Generating and Adapting Weights from Live Data (arxiv.org)

LLM Classification Is Feature Engineering | Minimally Sufficient (minimallysufficient.com)

For AI leaders Doom is a form of hype! - Erkan's Field Diary (erkansaka.net)

09.16

[2511.07885] Intelligence per Watt: Measuring Intelligence Efficiency of Local AI (arxiv.org)

Apple Reference Image: A New Approach for Verified Photography - Apple... (security.apple.com)

09.15

Introducing System One Models and Jev - TypeSafe AI Blog (typesafe.ai)

25 Years of Mass Surveillance Is Enough - Schneier on Security (www.schneier.com)

Inside the Inference Hardware Revolution Of 2026 - IEEE Spectrum (spectrum.ieee.org)

https://orib.dev/gefs.pdf (orib.dev)

09.14

[2609.12303] Breaking the Token Ceiling: Distilling Smaller, Stronger Byte Models (arxiv.org)

[2609.11873] The Last AI Built by Humans: Toward Genuine Recursive Self-Improvement (arxiv.org)

The Last AI Built by Humans: Toward Genuine Recursive Self-Improvement |... (academy.dair.ai)

Distributed Systems Classics (nvartolomei.com)

GitHub - rush86999/atom: Atom Agent, Open-Source Governed AI Agent Platform... (github.com/rush86999)

Why we built Pion | Andon Labs (andonlabs.com)

Foundation Model Engineering (sungeuns.github.io)

Claude Fable 5.1 Solves the Cyphral Distich (www.vals.ai)

09.13

Countering misuse of AI: September 2026 / Anthropic \ Anthropic (www.anthropic.com)

Getting 50 GB/s Back Out of the ANE | Eileen Yoon (eiln.github.io)

Measuring Reward-Seeking by Instilling Contrastive Beliefs (alignment.openai.com)

[2607.18966] Measuring Reward-Seeking via Contrastive Belief Updates (arxiv.org)

Houthis Used Claude Code to Develop Missile Guidance Software: Anthropic ·... (clashreport.com)

09.12

Real-SWE Benchmark — Specific Labs (withspecific.com)

AI researchers debate how close we are to recursive self-improvement (www.dwarkesh.com)

09.11

So you want to use OpenRouter? — Mo Moustafa (mmoustafa.com)

Declaration — Math and AI (mathandai.org)

Global Glacier Extinction Explorer (glacierextinction.com)

09.10

To write non-fiction, draw the trunk, then the rest of the tree (devz.cl)