John Builds

How do you keep your LinkedIn posts from sounding like everyone elses AI content?

by

There's a pattern on LinkedIn right now that's hard to miss. Every other post opens with a one-line hook, follows a numbered list, and ends with "What would you add?" You can tell within two sentences whether someone used AI or not.

The problem isn't AI writing — it's that most AI tools have no idea who you are. They generate from a blank slate every time.

I've been working on this problem with XreplyAI. The approach: train a voice profile from the user's own writing (tweet archive, past posts), then use that as the style layer before any generation happens. When you ask it to draft a LinkedIn post, it's not starting from scratch — it's working from your actual patterns.

We shipped LinkedIn scheduling recently. It supports text posts, images, and carousels. BYOK model so you're not paying a subscription AI cost on top of the tool.

Curious how others are handling this — do you use AI for LinkedIn content? What do you do to make it not sound generic?


https://xreplyai.com?utm_source=producthunt&utm_medium=social&utm_campaign=feature-2026-05-19

11 views

Add a comment

Replies

Best
George Sostak

For content creation I have a pipeline in Claude Code (through my custom version of Paperclip) that uses skill for every stage. /seo-content, /content-research, and /humanizer uses my voice-profile.md
Workflow A — Blog post (canonical):

Stage 0: /seo-pipeline           (pick from content/ideas/ or capture new)
Stage 1: /keyword-research       → GATE 1: approve targeting
Stage 2: /content-strategy        (format brief: category, word count, angle)
Stage 3: /content-research        → GATE 2: approve outline + sources
Stage 4: /seo-content + /humanizer (draft + strip AI patterns)
Stage 5: human editorial          → GATE 3: approve for publish
  5b: featured image + /image-optimize (1200x900, <300KB)
  5c: assign lead magnet (if any)
Publish: copy to site/content/posts/<slug>.md, push, Vercel auto-deploys
Stage 6: /seo-audit <live URL>
Stage 7: /content-atomizer        (2 LinkedIn + 1 thread + 2-3 tweets)