The 100-Hour Lesson: What Building with AI Really Takes cover art

The 100-Hour Lesson: What Building with AI Really Takes

The 100-Hour Lesson: What Building with AI Really Takes

Listen for free

View show details

About this listen

In this episode, Gil and I talk about where AI helps—and where it just frustrates the hell out of you. I share why I spent 20 hours trying to automate part of this podcast and then decided I didn’t even want to. Gil talks about switching from Cursor to Claude for code generation and what he’s learning from building his own AI assistant. Lots of honest chat about tools, burnout, and the emotional side of learning.

What we get into:


  • Why I scrapped my newsletter automation even though it technically worked
  • The tools we’ve been experimenting with lately: Claude, Cursor, Replit, Opus
  • Why your skills (and brain wiring) really matter when working with AI
  • How to spot when you’re trying to automate the wrong thing
  • Gil’s building an AI assistant—and has probably sunk over 100 hours into it
  • The problem with all those “1-click AI app” videos on social
  • How education (and parenting) might need to change in the AI era
  • Why it’s OK to suck at first—and why that’s part of the process

Notable quotes:


“I realised I don’t want to automate the newsletter—I want to write it.” – Nihal


“Even if what you built doesn’t ship, those 100 hours weren’t wasted—they were hours of learning.” – Nihal


“Just because it can be automated doesn’t mean it should.” – Gil


“If AI makes it too easy, we risk losing the grit and growth that comes from doing hard things.” – Gil


Tools mentioned:


  • Cursor
  • Claude
  • Replit
  • Riverside
  • Opus Clip


Connect with us:


Have AI questions? Connect with us on LinkedIn – we’d love to help you cut through the noise and start getting real value from AI.

  • Nihal Salah
  • Gil Chan


No reviews yet