The default AI LinkedIn post is cringe. You know it when you see it: line-broken-every-sentence, “Here is a hot take…”, a humble-brag personal anecdote, three bullet “lessons”, and a question nobody answers because the post sounds like a script. This tutorial replaces that template with a four-block frame (claim, evidence, tension, ask), an anti-cringe filter that catches the worst defaults, and a comment-engagement plan that turns LinkedIn from broadcast into conversation.
What this covers
A LinkedIn-native drafting workflow: claim-evidence-tension-ask block structure, anti-cringe filter (banned phrases, banned structures), formatting rules that respect feed pacing, and a comment-engagement plan you execute in the first two hours after posting. The output is a post that reads like a thoughtful peer wrote it, not an AI that learned from 2019 motivational posts.
Who this is for
Founders building a personal brand, B2B operators sharing playbooks, recruiters and hiring managers posting openings with context, and consultants using LinkedIn for inbound. Less useful for pure lead-gen accounts where the post is a thin wrapper around a CTA — those need a different (more direct) workflow.
When to reach for it
When you have a real point to make and 20-30 minutes. When you want to build a posting cadence (2-3 per week is the realistic floor). When you have a long-form piece — article, podcast, talk — and need a LinkedIn excerpt that earns clicks. Skip when the only goal is link distribution; LinkedIn punishes link-only posts.
Before you start
- Have one real point — a claim you would defend in a room. AI cannot manufacture conviction; it can structure one.
- Have evidence: a number, a story, a specific moment. Posts without evidence read as opinion noise.
- Know your audience. “B2B SaaS founders” is too broad. “Series-A SaaS founders who just hired their first head of marketing” lets AI calibrate.
- Reserve 2 hours after posting for comment engagement. The post is half the work.
Step by step
- Brief the AI with the four-block frame: CLAIM (one sentence, defensible), EVIDENCE (one specific data point or story), TENSION (what this contradicts or makes hard), ASK (one question that invites a story, not a yes / no).
- Generate 3 post variations. Different opening lines, same four blocks. Compare them side by side — the best one usually combines block 1 from variation A with block 3 from variation B.
- Run the anti-cringe filter. Remove: “Here is a hot take”, “Unpopular opinion”, “Most people get this wrong”, “I used to think… now I know”, any line ending in “Period.” Strip emoji bullets. Limit line breaks to between blocks, not between every sentence.
- Read the post as if a competitor wrote it. Would you respect them? If the answer is “this person is performing”, rewrite the weakest block.
- Format for feed pacing: claim as a one-liner with a deliberate gap below it. Evidence and tension as short paragraphs. Ask on its own line at the end.
- Plan the comment engagement: 5 prompts you will leave in your own comments to extend the conversation, and 5 people you will tag or reply to.
- Post and engage. First two hours after posting drive 80% of distribution. Reply within minutes, not hours.
First-run exercise
- Pick one real point you have actually defended in a recent conversation. Not a hypothetical thought-leadership topic.
- Run the four-block brief, generate 3 variations, combine the best blocks, run the anti-cringe filter.
- Post at your usual posting time. Set a 2-hour timer for active comment engagement.
- After 24 hours, check the ratio of comments to likes. Above 1:10 means the ask worked.
Quality check
- Does the claim survive a “so what” test? If your reader thinks “okay, and?” the claim is too soft.
- Is there one specific number or moment? Vague evidence reads as filler.
- Does the tension feel real? “Most people think X, but actually Y” is fake tension if X is a strawman. Steelman the opposing view.
- Is the ask answerable in 2 sentences? Long-answer asks die in the feed.
- Strip every line that does not move the four blocks forward. LinkedIn rewards short.
How to reuse this workflow
- Save the four-block frame as a doc. The prompt does not change; the input does.
- Build an anti-cringe filter list specific to your voice. Add new banned phrases as you spot them.
- Track which claims drive comments vs. likes. Comments compound; likes do not.
- Re-audit posting cadence every month. 3 strong posts beats 7 mediocre.
Recommended workflow
One real claim -> evidence + tension + ask in four-block brief -> 3 AI variations -> combine best blocks -> anti-cringe filter -> format for feed pacing -> plan comment engagement -> post -> active engagement first 2 hours -> review next morning.
Common mistakes
- Posting without a real claim. AI cannot manufacture conviction.
- Using AI’s default formatting. Line breaks between every sentence signals AI-generated.
- Cringe phrases (“Here is a hot take…”). Banned, no exceptions.
- Fake tension. Strawman opposing views read as low-effort.
- Posting and disappearing. Comment engagement in the first 2 hours is the distribution lever.
- Asks that are yes / no questions. Open-ended asks pull comments.
FAQ
- How often should I post?: 2-3 times per week sustainably beats 7 times for 3 weeks then nothing.
- Should I use hashtags?: 3 relevant ones, not 10. The over-hashtagged look reads as 2018 LinkedIn.
- Can I let AI write the whole post?: No. Let AI structure your point. The point comes from you.
- What about LinkedIn newsletters?: Different workflow. Newsletters need a series arc; this template is for single posts.
- How long should the post be?: 800-1,500 characters. The “see more” cliff is at 210 characters — your hook must work in the first 2 lines.
Related
- AI personal branding workflow
- LinkedIn lead magnet tutorial
- AI content marketing stack guide
- AI newsletter tutorial