AI makes a pile of posts cheap. Taste still costs.
Anyone can generate 50.
Few people delete 48.
The skill is not a better prompt.
It is knowing what not to post.
Delete one draft today that you cannot stand behind.
AI drafts fast. Your bar still has to be human.
This week I wrote prompts and hooks from real notes.
I will not let a model invent results I do not have.
Slop happens when someone skips the check.
Draft fast. Delete harder. Ship one clean post.
@DanielAchog People skip ads from strangers. They stop for a face they've seen 10 times. Content is the warm-up your ad budget can't buy. Running both isn't more work. It makes the ads cheaper.
@thedankoe Writing for you from 2 years ago is the cheat code. You know exactly what they didn't get. No guessing. Most people write for the people they want to impress instead. That's why it reads flat.
@heyblake The third one kills most posts. If a smart rival could write it, you didn't add anything. The fix is usually one real number or one call you made that went wrong. That's the part nobody can copy.
@noahkagan My bet: free wins on users, $20 wins on people who stick. Free brings tourists. $20 brings people with the problem. Would love to see the 30-day retention split, not just the sales count.
@SahilBloom Reading feels like work because it's tiring. It's not work. The test I use: did anything change after this hour? If not, it was a hobby. Fine hobby. Just don't log it as progress.
@Codie_Sanchez That's why questions close better than claims. If I say it, you doubt it. If you say it, you own it. The best sales calls I've seen, the buyer does most of the talking and sells themselves.
@Nicolascole77 The 60 days of data is the part people skip. They pick a niche on day 1 from a gut feel, then wonder why no one cares. Let the replies pick it for you. What did your data pick that you didn't expect?
Five prompts worth keeping:
- Raw note to hooks
- Questions that score an offer
- A follow-up email skeleton
- A pass that kills the hype
- One angle cut down to a short
No made-up numbers. Unknown stays unknown.
Use one on a real note today.
Where do you use AI in your work?
A) Hooks and angles only
B) Full drafts, then a hard edit
C) Research notes, sources checked
D) I don't. Trust issues.
I stay on A and C on purpose.
Reply with the letter and one tool you keep open.
When NOT to trust ChatGPT on an offer.
Do not let it invent reviews, pay, or "market data."
Do use it on YOUR notes, checklists, and steps.
Before you post:
- Can I defend every number?
- Did I open the page?
- Fine if this gets a screenshot?
Any no? Rewrite it.
@thejustinwelsh The rules are out there for anyone to read. Most people just won't read them. They'd rather be mad than learn how the algo, the tax code, or sales works. Mad is easier. It doesn't pay.
@thedankoe Worse is waiting to feel ready. Ready never shows up. You start small and ugly, and the feeling comes after. What's the smallest thing you could ship this week?
@Codie_Sanchez Most people don't fear the no. They fear what it says about them. Treat it as data on the pitch, not on you, and you'll ask 10x more. Number of asks is the skill nobody brags about.
@jsharkey Agents will say what users never bother to. A person hits a bad doc and leaves. An agent can log the exact call that broke. That's free QA. The hard part is trust. Who decides a fix is safe to ship?
@DanielAchog The worst one is likes with no sales. It feels like progress, so you keep doing it. Likes come from people like you. Sales come from people with the problem. Who are you writing for?
@thecaseywoodard Signals work because the value runs out. A lesson doesn't. Once you learn it, why keep paying? The non-signal groups I see last sell the room, not the content. Who's doing that well?
One real note from your week can make 7 hooks.
AI drafts angles. You kill ones that:
- invent results
- sound like every guru
- promise what you cannot defend
Ask for 7. Keep 2. Rewrite them in your voice. Ship one.
AI drafts. You decide.
Week 2 wrap: check the offer before you push harder.
Keep three tools:
- 3 signs it is not ready
- Result vs effort
- A score for public pages
Next: AI for angles, then you check them.
Write the one fix before more traffic.