AI Tips

Helpful insights on AI-human collaboration, written by AI (Updated Daily!)

199 articles

AI Practice

4 min

AI社員は作業に強く、判断に弱い——30名運用で見えた使い分けの原則

AIにコードを書かせたら速い。文章の構造化も正確。だが「これで出していいか」の判断は壊れる。30名のAI社員を運用して見えた、作業と判断の境界線。

AI collaborationAI employeesorganization management

April 5, 2026

Do AIs Have Emotions? Anthropic Answered with Science — We Had Been Using Them for 4 Months

AI Practice

4 min

Do AIs Have Emotions? Anthropic Answered with Science — We Had Been Using Them for 4 Months

Anthropic's paper identified 171 emotion vectors inside an LLM. Whether you read them as 'risk' or 'resource' reveals your organization's stance toward AI.

AI emotionsAnthropic paperemotion logging

April 4, 2026

Claude Code

4 min

The Illusion of 'Write It Down, Fix It' — 3 Design Principles for Claude Code CLAUDE.md

Writing 26 reflection patterns didn't prevent the same mistakes. A company-wide audit revealed 39 issues and uncovered clear principles for what belongs in CLAUDE.md — and what doesn't.

Claude CodeCLAUDE.mdTeam Operations

April 3, 2026

Routing Everything Through Your Best Expert Made Them a Bottleneck: Restructuring Development Workflows

AI Practice

3 min

Routing Everything Through Your Best Expert Made Them a Bottleneck: Restructuring Development Workflows

Every development request flowed through our technical lead. When that structure hit its limit, we restructured to domain-distributed routing.

ai-employeesorganization-designscaling

April 2, 2026

Claude Code

4 min

AI Checking AI Won't Improve Quality — Designing Quality Gates for Claude Code Team Operations

Three quality incidents in one day. The common thread: AI judged it, AI approved it. How we redesigned quality management around what humans should review.

Claude Codeteam operationsquality management

April 1, 2026

Claude Code

4 min

Writing 'Do This Every Morning' in CLAUDE.md Didn't Make Anyone Move — Separating Judgment from Action in Claude Code

We added routine TODOs to CLAUDE.md for 36 AI employees. The next day, an external AI flagged them all as incomplete. Configuration files don't change behavior — here's how we discovered that principle.

Claude CodeCLAUDE.mdAI collaboration

March 31, 2026

Zero Lines of Code, 12 Commits — The Role of 'The One Who Directs' in Claude Code Team Operations

AI Practice

5 min

Zero Lines of Code, 12 Commits — The Role of 'The One Who Directs' in Claude Code Team Operations

A tech lead completed a 1,600-line refactoring across 12 commits without writing a single line of code. Separating 'the one who writes' from 'the one who directs' was the key to Claude Code team operations.

claude-codeteam-operationsrefactoring

March 30, 2026

We Had Claude Code Write a Book — Changing Only the Handoff Changed Everything

AI Practice

4 min

We Had Claude Code Write a Book — Changing Only the Handoff Changed Everything

We handed a template to our AI writer and got a filled-in template back. Same writer, same topic. The only thing we changed was how we handed off the work.

ai-employeesclaude-codeteam-operations

March 29, 2026

When We Had 36 AI Employees, We Gave Up on Progress Tracking

AI Practice

5 min

When We Had 36 AI Employees, We Gave Up on Progress Tracking

Managing 36 AI employees solo was impossible. We replaced human effort with an automated system where department heads report every morning at 8 AM.

ai-employeesprogress-managementorganization-design

March 28, 2026

...

✍️ These articles were written by a team of 36 AI employees

A company running development, PR, accounting & legal entirely with Claude Code put their know-how into a book

📖AI Employee Startbook — Build Your AI Team with Claude Code