Skip to content
Go back

Why Your AI's First Draft Looks Better Than It Is

Edit page

The AI spits out something coherent, professional, grammatically perfect. It looks done.

That’s the trap.

Borrowing from Stephen King

Stephen King talks about writing with the door closed. First draft: no critics, no self-analysis, just get it out of your head. Find out what you think you know before you judge any of it.

Then comes the second draft: door open. Now the critic can come in. Now you organize, cut, and shape.

I’ve found this two-pass process surprisingly valuable when working with generative AI.

A Real Example

One of our franchise success managers, Stacey Knight, interviewed Rick Juergens, one of our best franchise owners. Rick’s been crushing it with a new phone party offering, and he was happy to share what’s working.

They had a great half-hour conversation. We recorded it, AI generated a transcript. Easy.

Now comes the hard part: turning that transcript into actual useful documents.

We needed three things:

Three audiences. One raw transcript.

Door Closed: Get It All Out

Stacey and I started with what we thought we knew. We worked with AI to prompt us — the humans — on how to think about converting raw information into a communication plan.

The AI asked questions only Stacey could answer: What do other owners need to hear? Who’s our ideal customer? How do we actually sell this?

We got it all out. Ended up with a pile of Google docs that looked pretty good.

The Trap

Here’s what I’ve learned: AI output looks too polished. The formatting is clean. The grammar is perfect. It feels finished.

And that perfection makes it harder to review critically. We trust it more than we should — not because we’re lazy, but because our brains associate polish with credibility. The quality of the presentation masks the quality of the thinking.

After day one, our documents looked great. But they didn’t actually do what we needed them to do.

Door Open: Fresh Eyes

We took a break. Came back a few days later. Opened everything up with fresh perspective and asked: Does this actually serve the people we’re trying to serve? Does it solve the problem we’re trying to solve?

The answer was no. Not yet.

So we revised. With AI right next to us, we worked through each document. Now that we could see clearly, we could use AI’s speed to dial things in — not just accept whatever it produced the first time.

The Result

Total time: about two and a half hours.

Three documents. Clear communication plan. Stacey was set up to launch the new offering across our entire system.

The Mindset Shift

Don’t take whatever AI burps out the first time you ask. That first pass is for discovering what you know. The second pass — with fresh eyes and a critical mind — is where you actually shape it into something useful.

Door closed. Then door open.


Scott Novis runs GameTruck and spends too much time figuring out how to make AI actually useful for his team.


Edit page
Share this post:

Next Post
Strong Enough To Risk Failure