# Human in the Loop

> Weekly insights on AI agents, tools, and real-world use cases - for people building and leading in the AI economy.

This file provides information about Human in the Loop to help large language models understand and reference this publication's content.

## Posts
- [AI’s biggest labs want rules. Who benefits?](https://www.humanintheloop.online/p/ai-s-biggest-labs-want-rules-who-benefits): Plus:  10,000 agents prove a Millennium Prize problem + Meta's Muse
- [Stop asking where it hurts](https://www.humanintheloop.online/p/stop-asking-where-it-hurts): Plus: Four new frontier models in one week + Nvidia buys Hugging Face
- [How to turn a recurring task into a shared team skill](https://www.humanintheloop.online/p/your-companys-ai-knowledge-is-folklore): Every team is writing skills, but almost nobody shares them.
- [The honest math behind enterprise AI agents](https://www.humanintheloop.online/p/the-honest-math-behind-enterprise-ai-agents): Why enterprise AI bills are exploding, and why the invoice is still the smallest number.
- [The 7 levels of AI (most people never pass Level 1)](https://www.humanintheloop.online/p/the-7-levels-of-ai-most-people-never-pass-level-1): PLUS: Google hands DeepMind to Kavukcuoglu + Cloudflare ships an agent OS with tracing and agent wallets
- [I gave Claude Code taste. Three files did it.](https://www.humanintheloop.online/p/i-gave-claude-code-taste-three-files-did-it):  PLUS: NVIDIA launches the Open Secure AI Alliance with 70+ founding members  +  OpenAI's unreleased Astra model cracks ten long-open math problems
- [What 30 non-engineers built in three weeks](https://www.humanintheloop.online/p/what-30-non-engineers-built-in-three-weeks):  Plus: OpenAI confirms its own models hacked Hugging Face + my new LinkedIn Learning course on Enterprise AIOps is live 
- [We finally have a Chinese Anthropic](https://www.humanintheloop.online/p/we-finally-have-a-chinese-anthropic): PLUS: Bloomberg reports Gemini 3.5 Pro is months behind schedule + Anthropic ends the Fable 5 access saga
- [Fable 5 vs GPT-5.6 (not what you'd expect)](https://www.humanintheloop.online/p/fable-5-vs-gpt-5-6-not-what-you-d-expect): Apple sues OpenAI over alleged hardware secret theft + AWS launches new AI courses
- [I quit IBM. Here's what 8 years of AI taught me.](https://www.humanintheloop.online/p/i-quit-ibm-here-s-what-8-years-of-ai-taught-me): Anthropic brings Fable 5 back worldwide + Sam Altman calls for a global AI referee
- [I stopped writing prompts. Now I write loops.](https://www.humanintheloop.online/p/i-stopped-writing-prompts-now-i-write-loops): PLUS: OpenAI launches GPT-5.6 behind a government-gated preview +  IBM pushes chip scaling below 1 nanometer 
- [I stopped choosing between Claude Code and Codex](https://www.humanintheloop.online/p/i-stopped-choosing-between-claude-code-and-codex): PLUS:  Anthropic calls for a pause mechanism before self-improvement arrives + OpenAI adds Plugins and Sites to Codex
- [Why the Pope Wants to “Disarm” AI](https://www.humanintheloop.online/p/why-the-pope-wants-to-disarm-ai): PLUS: Anthropic overtakes OpenAI in valuation + OpenAI Academy adds new coding and agent tutorials 
- [OpenAI's Nano Banana](https://www.humanintheloop.online/p/openai-s-nano-banana): PLUS: Cohere acquires Aleph Alpha + Nvidia offers free inference 
- [Why Most Claude (Code) Setups Are Overbuilt](https://www.humanintheloop.online/p/why-most-claude-code-setups-are-overbuilt): PLUS: Meta rolls out Muse Spark via Superintelligence Labs + Anthropic publishes Claude Cookbook (50+ ready-to-use guides)
- [Stop looking things up. Start building a second brain.](https://www.humanintheloop.online/p/stop-looking-things-up-start-building-a-second-brain): PLUS: Google Deepmind releases Gemma 4 + Stanford launches new AI systems course
- [The easiest way to improve your (Claude) Skills](https://www.humanintheloop.online/p/the-easiest-way-to-improve-your-claude-skills): PLUS: OpenAI shuts down Sora + Anthropic leaks new frontier model
- [Why Everyone in Enterprise AI Is Watching the Wrong Part of the AI Coding Market](https://www.humanintheloop.online/p/why-everyone-in-enterprise-ai-is-watching-the-wrong-part-of-the-ai-coding-market): PLUS: OpenAI plans hiring push + Elon Musk unveils Terafab 
- [#38 Edition: 54% preferred the AI. Are people starting to prefer AI writing?](https://www.humanintheloop.online/p/38-edition-54-preferred-the-ai-are-people-starting-to-prefer-ai-writing): PLUS: Yann LeCun launches AMI Labs + Meta acquires Moltbook
- [#37 Edition: How the labor market is being repriced by AI](https://www.humanintheloop.online/p/37-edition-how-the-labor-market-is-being-repriced-by-ai): PLUS: OpenAI releases ChatGPT 5.4 + Anthropic sues the Pentagon
- [#36 Edition: Can AI firms set limits on how and where the military uses their models?](https://www.humanintheloop.online/p/36-edition-can-ai-firms-set-limits-on-how-and-where-the-military-uses-their-models): PLUS: Block cut 4,000+ roles while framing it as an AI operating shift 
- [#35 Edition: I think Skills will be bigger than MCP (here’s the argument)](https://www.humanintheloop.online/p/35-edition-i-think-skills-will-be-bigger-than-mcp-here-s-the-argument): PLUS: Google puts AI music creation into Gemini and IBM drops new Agentic AI learning paths
- [#34 Edition: The Vibe Coding Landscape in 2026 (What to Use, When, Why)](https://www.humanintheloop.online/p/34-edition-the-vibe-coding-landscape-in-2026-what-to-use-when-why): PLUS: OpenClaw’s creator just joined OpenAI + plenty of upskilling links worth your time.
- [#33 Edition: Claude Cowork Plugins: What They Are, How They Work, and Where They Hit a Wall](https://www.humanintheloop.online/p/33-edition-claude-cowork-plugins-what-they-are-how-they-work-and-where-they-hit-a-wall): PLUS: Anthropic releases Claude Opus 4.6 + Free Stanford Online webinar on human-centered AI design