A national sales manager just spent a month building a custom CRM with AI. It's very good. That's the problem.

Alex Hormozi made the case on a recent episode, and it's the sharpest framing I've heard: the great misconception is that AI somehow cancelled every other form of leverage.

It didn't. Capital still compounds. Media still scales. A good offer still outsells a bad one with better technology behind it. Teams still multiply output. AI joined the list of levers — it didn't replace the list. And here's his kicker, the one I'd paint on the office wall:

Read that twice. It's the whole disease in one sentence.

Here's the mechanism, because it's sneaky. AI massively increases your capacity to do things. So you start doing things you would never have done before.

But think about why you'd never have done them. They were at the bottom of the list. They were at the bottom because they didn't matter enough. And now — with infinite cheap capacity — the bottom of the list is getting done. Faster! Automated! Beautifully!

My sales manager mate didn't build a CRM because his business needed one. He built it because, for the first time, he could. AI didn't give him leverage. It gave him a power tool and pointed it at his hobby.

Scarcity used to do your prioritising for you. When you couldn't do everything, you were forced to do the thing that mattered. AI quietly removed the forcing function — and most businesses haven't replaced it with anything.

One question, and it's brutal because it's binary:

Every business has one constraint — the bottleneck that's actually holding revenue back. Not five. One at a time. Not enough demand, or not enough of that demand converting, or delivery falling over when both of those work.

Everything pointed at the constraint moves the needle. Everything pointed anywhere else is decoration — automated decoration now, which is worse, because it feels like progress and produces reports.

And some of the highest-leverage fixes need no AI at all. One-to-one delivery becomes one-to-many: 10x, one decision. A sales motion that educates prospects before the call, so two reps close what used to take ten. Killing a project that stopped mattering — that one's free, and it pays instantly. The fundamentals didn't retire when the robots arrived. Hormozi's wealthiest mates barely touch AI personally. Their teams use it — pointed at constraints. That's the tell.

So where does AI actually belong? Where the constraint usually lives and the dollars are countable. In almost every business we audit, that's the same short list.

Speed to lead — an enquiry answered in 30 seconds instead of four hours, nights and weekends, because after five minutes your odds of reaching them drop ~10x. The dormant database — the thousands of contacts you already paid for, worked back into booked calls one conversation at a time; we pulled $400K a month out of one list everyone had written off. And your ad signals — feeding settled revenue back to Google and Meta so the machine hunts buyers, not browsers.

Notice what those three have in common: each one ends in a booked call or a banked dollar. Not a dashboard. Not a workflow. A calendar with names in it.

Notice also what's not on the list: custom CRMs, meeting summaries, automated internal emails, the org's fourth content repurposing pipeline. Fine tools. Bottom-of-the-list tools. The stuff scarcity used to protect you from.

To be fair to my sales manager — and I say this with love, because his CRM really is very good — the instinct underneath it is right. He could see AI was a big deal and he wanted his hands on it. The instinct isn't the problem. The target is.

Same bloke, same energy, same month — pointed at his actual constraint? An agent working his stalled pipeline one conversation at a time, every quote followed up five times instead of 1.7, every dead deal from the last two years asked one good question. That month makes him the best-performing sales manager in the country. The CRM makes him... organised.

The gap between those two months is the entire AI economy right now.