A leadership workshop on delegating to AI stalled on a hard question; the instructor let AI act on the room, then asked it how to fix the workshop, and the session ended on an untested experiment for knowing when to stop AI.
Summary: A leadership workshop on delegating to AI stalled on a hard question; the instructor let AI act on the room, then asked it how to fix the workshop, and the session ended on an untested experiment for knowing when to stop AI.
I challenged several LLMs to generate funny Tamil puns ending in ".ai". While DeepSeek and Claude struggled with accuracy, Gemini captured the cultural nuance perfectly with clever wordplay like Tholl.ai (annoyance) and Kaval.ai (worry).
AI performance depends less on clever prompts than on context. Give models the right goals, examples, constraints, history, and tools so they can act effectively.
I used the Claude Code agent to search 24,000 Wikipedia entries for names starting and ending with "AI." In ten minutes, I identified ten individuals, including Aishwarya Rai and Ai Nagai, through automated pattern matching and scripting.
I learned how Substack RSS feeds can reveal my reading, why Claude answers more briefly on mobile, how Cloudflare enables agent trial accounts, and how Git can ignore files outside .gitignore.
