Why generic AI-written posts fail, and how to write with a stance

Generic AI posts fail in a way that is easy to feel and annoying to name. They are grammatical. They are often correct. They still read like a brochure afraid of the reader. Nobody replies, because there is nothing to reply to. A stance is the missing object: who the post is for, what you believe, and what that person gains or loses.

What generic AI posts sound like#

It opens soft, names a topic like pricing or focus, lists two benefits and two caveats, and closes with a proverb or "what do you think?" Nothing in it could embarrass the author or help a specific person decide. The model averaged every acceptable opinion. Averages get a polite like, not a reply.

This is not a reason to avoid a model. It is a reason to reject the draft that has no one in it. A coach is useful after a stance exists. It is useless when you ask it to invent the opinion and you ship whatever comes back.

Four ways a draft dies#

The hedge#

The hedge is a claim with the claim removed.

"Pricing is important, but it depends on your stage, your market, and your customers, so do what feels right."

Every clause after "but" is an exit. A stance picks the case: a public price is how a stranger knows the conversation is for them. Or the opposite, if you believe the opposite. A post can survive one sharp sentence. It cannot survive a paragraph of insurance.

The proverb#

The proverb is the last line that turns a scene into a poster. "The lesson is to talk to users." "Remember, consistency beats intensity." They announce that the post is finished and the reader is a student. People reply when you stop on the cost, the choice, or the ask. Cut the proverb. If the post no longer says anything, the middle was empty.

The scene that ends in a command#

"A founder stares at the pricing page for a week. The button still says contact us. Name the price before you write another feature thread."

The scene is specific. The ending is a lecture. A stance does not need to give an order. It needs to show who pays: the stranger who cannot tell if the product is for them. End on the cost, the belief, or a constrained ask. Those asks are the reply-driving formats.

The essay question#

"What are your thoughts on AI?" is not a stance and it is not an ask. It asks the reader to do your thinking in public. A question mark only looks like engagement. The reader has to choose the topic, the take, and the length. They will not.

A real ask has a constraint. "Name one workflow you refused to hand to a model. Not email." That is a belief — some work should stay manual — dressed as a question a person can answer before their coffee cools.

A stance is a reader and a cost#

You do not need a brand voice document. You need three lines, written before the draft, that the draft is not allowed to betray.

Who is this for? Not "founders." A person you can picture: an indie builder with a small account, a product that exists, and a habit of posting when they feel guilty.

What do you believe? One sentence you would still say if a peer pushed back. "A small account grows from replies it can predict, not from announcements it hopes will travel."

What do they gain or lose? The afternoon lost to another vague draft. Or a post a peer can answer in one line, if they take the belief seriously.

If you cannot fill those lines, you have a topic, not a post. Topics are how generic drafts begin.

A reply-style post may skip the bet when the ask is doing the work. "Name one, not the obvious one" does not need a manifesto. It still needs a reader who recognizes the constraint as theirs. The shape is not the stance. The reader is.

Write the belief first#

Open the composer only after the belief is a sentence you can see. Then say that sentence to one person, with the cost still attached.

Here is the same topic twice. Neither example is a quote from a real account. They are the edit, so you can see it.

Generic: "Building in public can be a powerful way to grow. Share your journey, be authentic, and stay consistent. There are pros and cons, so find what works for you. What are your thoughts?"

With a stance: "Building in public fails when you post the milestone and hide the offer. The people who can buy do not need your follower count. They need the sentence that says who the product is for. Post that sentence before you post another screenshot of the dashboard."

The second one can be wrong. Good. A peer can answer with the week the screenshots actually started the conversations. You can answer that. You cannot answer "so true." Write the belief in words you would say out loud, and delete any line you would not say to a peer.

An edit that keeps the stance#

Mark every hedge and keep one only when you name the tradeoff. "Charge less and you need more customers. Charge more and you need a clearer sentence." is a tradeoff. "It depends on your goals" is a hallway.

Delete the proverb and the closing command when the post still stands without them. If it collapses, the stance was only in the last line. Move the belief to the front.

Replace "everyone" with the person from your three lines. "Everyone should post daily" is a poster. "If you post when you feel guilty, four slots beat a guilty thread" is a stance, and the daily routine is that claim as a calendar.

Cut press-release lines, including ones for Xcend. "Use an AI coach" is not a stance. "Do not ship a draft that will not risk a disagreement" is.

Then read it as the person you named. Could they answer in a line, or tell you that you are wrong with a scene from their week? If not, it is still generic. A streak of hedges trains the account to sound like everyone else.

Where a coach helps, and where it does not#

A model is good at speed, variation, and catching a sentence that wandered. It is bad at caring whether you believe the sentence. Ask it for "a thread about consistency" and you will get the proverb. Hand it the reader, the belief, and the cost, and reject drafts that sand those lines down, and you get something you can edit into your own mouth.

That is the order Xcend's coach is meant to follow: a stance and a shape first, not a topic cloud. The dangerous draft is the flattering one. It sounds professional, and next week the account still has no position anyone can find. Put one reader in the chair, say what they lose, and delete every line that tries to be liked by the rest of the room.