LinkedIn content engine

The problem

Posting consistent, on-brand content on LinkedIn is a known growth lever — and a notorious time sink. The hard part is not the writing itself; it is the constant context-switching between catching an idea, structuring it, drafting it in a voice that actually sounds like you, and getting it over the finish line. Most attempts at “AI for content” collapse it all into one prompt and produce generic posts that drift from your voice within a week.

What I built

A content workflow that splits the job into discrete, repeatable steps — each one handled by its own Claude Skill, each one small enough to be reliable in isolation.

• Idea capture. Voice notes or one-liners go in; a structured idea brief comes out.
• Angle & structure. The brief is shaped into a post outline with hook, body and call-to-action.
• Draft & tone. A first draft is generated in my actual voice, calibrated against my back catalogue of published posts.
• Polish loop. A tight back-and-forth until the post is ready to publish.

Because each step is its own Skill, the pipeline is predictable and re-runnable. If one stage drifts, I fix that stage — not the whole prompt.

The result

In daily use for my own LinkedIn output. The same pattern is offered to clients who want a content engine without becoming prompt engineers themselves.