A hands-on introduction to Claude Code. Hosted by Paw.
Everyone has tried AI. Almost nobody puts in the 20 hours it takes to get properly good at it.
Refusing AI in 2026 is insisting on paper mail because you don't trust email.
Most people's mental model of AI is 3+ years old.
> I want to clean up and organize my Downloads folder. First look at what's in there and ask me clarifying questions before doing anything. Then propose a plan, and only execute after I approve it.
Anything unclear so far? Ask now.
Every session, it wakes up fresh. So write better notes for it to wake up to.
> Which emails from the last 3 days actually need a reply from me? One line each, ranked by urgency.
> Look at next week's calendar and write my Monday-morning briefing: what's coming, what needs prep, where am I double-booked?
> Find the oldest email still waiting on me and draft a reply in my tone. Show me. Don't send.
Questions before we break for lunch?
"I violated every principle I was given: I guessed instead of verifying."
Last questions before you build your own thing?
> I want to [your project]. I'm not technical, so make this as easy as possible for me. Ask me clarifying questions first, then give me a step-by-step plan. Whenever you need something from me (a login, a click, a decision), give me exact instructions.
AI takes the parts that were never the joy.