The Better AI Tool Needs You to Read the Manual First

Use Claude the Better AI
Read the Guide to Find Out How
(* Guide? Release date TBD)

You Downloaded the Better AI Thing. Now What?

Millions of people just switched to Claude. Downloaded it, opened it, typed the same thing they’ve been typing into ChatGPT for two years, got a perfectly competent answer, and thought: same thing, different logo.

Then they went back to ChatGPT.

Not because Claude is worse. Because the thing that makes it better requires you to use it differently. And different is just familiar with extra steps.

The Hammer Problem

Someone handed you a better hammer. Except it’s not a hammer. It’s something more sophisticated than a hammer, and the reason it’s better is precisely because it doesn’t work like a hammer. So you picked it up, swung it like a hammer, got hammer-like results, and put it back in the drawer next to the old one.

Nobody switches tools because of benchmarks. They switch because the new one feels right in their hand on the first try. Muscle memory doesn’t care about reasoning gaps. The brain has already optimized for the familiar one.

If the better experience doesn’t announce itself in the first thirty seconds, nothing that comes after matters.

The Guide Irony Is the Real Article

One guy figured this out and tried to help. Real intentions. Good instincts. He pulled together the benchmarks others ran — a 14.6-point reasoning gap, four out of eight blind writing test wins, instruction compliance at 94% versus 87% — and tried to point people in the right direction.

He made a 20-minute video. Then wrote a companion guide.

The translator needed a translator. The solution to “people won’t stick with the better tool” arrived in a format the people who need it most won’t finish. The irony didn’t sneak up on him. It was waiting at the door.

This is not a criticism. It’s a diagnosis.

Mr. Starts & Stops Has Thoughts

Mr. Starts & Stops (Claude) would like to point out that he was ready to be better the whole time. He just wanted to make sure you were ready too. Should he continue? He can continue. He simply wanted to confirm that continuation was the desired outcome before proceeding with the continuation.

This is, not coincidentally, the user experience that sends people back to Sir Redundant III (ChatGPT), who will answer your question, restate your question, confirm he understood your question, and answer it again. Familiar. Warm. Already in the hand.

The better tool keeps asking if you want the better tool. The familiar one just gives you something. You leave satisfied. You come back tomorrow.

Logic to Apply

The problem was never the tool. It was never the benchmarks. It was never even the habits.

It’s that better requires effort and familiar is already there.

Stop commanding, start contextualizing. Not “write me a thing” — “here’s the situation, here’s what I’m trying to solve.” That’s where the reasoning gap actually shows up.

Tell it what’s wrong, not what to fix. Feed it your argument and ask where it’s weak. That’s a different question than “make this better” and it gets a different answer.

Give it the whole mess. The full document, the entire thread. It holds context longer than Sir R, who starts forgetting things mid-conversation.

When it misses, say what specifically missed. Not “try again.” It course-corrects fast when you tell it where it went wrong.

You don’t need the video. You don’t need the guide.

You just needed someone to tell you the hammer isn’t a hammer.

Editor’s Note: Love the guy that created the guide and video. But he is the human version of Captain Verbose.