I must start from this:
I resisted AI for months.
Not because I'm a Luddite. I've been testing productivity tools since I started with online business. I resisted because I'd seen the output. Generic corporate slop. Soulless garbage flooding Amazon.
I thought AI was for people who didn't care about quality.
Then I ran a 72-hour experiment that changed my mind.
I — The test
Same book concept (sourdough bread guide). Same quality bar. Two methods.
- Method A (traditional): 22 hours over 14 days. 8,500 words. Quality: 7/10.
- Method B (AI-assisted): 6 hours over 3 days. 9,200 words. Quality: 7.5/10.
Better output. One-quarter of the time. Same person doing the work.
What changed wasn't the writing. It was where I spent my brain.
Traditional approach: 80% typing. 20% thinking.
​AI approach: 20% typing. 80% thinking.
This wasn't working less. It was working at a higher level.
II — The pattern I wasn't expecting
Over the next 8 months, I tested this with hundreds of publishers.
The pattern was identical across all of them:
- Week 1: Terrible results (everyone wants to quit here)
- Week 2: First usable output (60% good, 40% needs work)
- Week 3: Inflection point—suddenly it clicks
- Week 4+: Operating at 3-5x previous speed with equal or better quality
The ones who succeeded pushed through week one.
The ones who failed quit on day three.
Question: What actually happens when you learn this?
You stop being a typist. You become a director. Your job shifts from "putting words on page" to "deciding which words belong on the page." Spielberg doesn't operate the camera. Gehry doesn't lay bricks. That's not where the value lives. The value is in the decisions.
III — What should concern you:
This is a scenario of 2 publishers over 36 months. Both skilled. Both disciplined.
Publisher A (traditional method):
- Output: 4 books/year
- 36-month total: 12 books
- Revenue: $89,000
Publisher B (AI-assisted method):
- Output: 15 books/year
- 36-month total: 45 books
- Revenue: $267,000
Same standards. Same talent. 3x the output.
But here's what matters more than revenue: Publisher B ran 3x more experiments. Gathered 3x more data. Learned 3x faster. The gap compounds. After 5 years, it's not 3x. It's 10x. Then 20x.
The Uncomfortable Question:
What happens when someone with your exact skill level learns AI, and you don't?
They don't replace you. They just move faster. Test more. Publish more. Learn more. You don't disappear overnight. You just become... slower. Quieter. Less relevant. Not because you can't do quality work. But because the leverage gap becomes unbridgeable.
IV — What I'd do if starting today
5-day test. Treat it like learning any skill.
- Day 1: Use AI for something small daily (email drafts, social posts)
- Day 2: Generate one first draft. Mark what sucks. Learn from it.
- Day 3: Focus on prompt quality (specific instructions beat vague ones 10:1)
- Days 4-5: Complete one full project. AI drafts. You direct, refine, decide.
If it still feels useless after 5 days, you tested it properly. Walk away with data.
But I'll bet you $100 that by day 3, you'll have at least one "holy shit" moment where you realize the leverage potential.
The End
The future isn't "AI vs. humans."
It's "humans who learned AI vs. humans who didn't."
The first group is already compounding. The second group is still debating whether it's "real work." You get to choose which group you're in. But only for the next 12-24 months. After that, the gap gets too wide.
— Nic
See our Programs:
​Follow us on X/Twitter​