[Watch the video of this essay here.]

AI has made a generation of founders believe they should do everything themselves.

Design the pitch deck. Write the copy. Build the landing page. Edit the video. Why pay someone when Claude can help you figure it out?

Here's what that logic misses: some moments don't offer a learning curve.

You don't get to "iterate" on your Series A pitch while Sequoia waits patiently. You can't A/B test your product launch keynote in front of 2,000 people. Your first sales call with the dream client isn't a sandbox environment.

There's a difference between tasks where "good enough" works and moments where the gap between amateur and professional is everything.

I've watched founders spend weeks using AI to build a pitch deck, then walk into the room and forget that they still have to deliver it. Whoops! Or while they were dinking around with AI they forgot they still have a product to build. The slides were fine. The story was scattered. The nerves were obvious. The check didn't come.

I've spent 20 years in those rooms. I’ve helped a team win TechCrunch Disrupt. I’ve coached Harvard hackathon finalists hours before they took the stage to win. I’ve worked with founders who quickly raised tens of millions.

Not because I'm smarter than AI. But because I've already made the mistakes they're about to make - just not in front of the people who write checks.

And side note, yes, I am smarter than AI, obviously, as this picture will show.

Now AI is an incredible tool for the 90% of work where repetition and refinement are possible, as I’ll show you more and more shortly.

But for the 10% where you have to get it right the first time? You need someone who's felt that specific pressure. Who knows what actually matters when the lights come on.

The question isn't "can AI help me do this?"

It's "can I afford to learn this lesson in front of a live audience?"

Take it from one overconfident amateur to another:

Sometimes, one chance is all you get.