The short answer: I use Claude for every stage of content creation — from initial topic research and outline structure, through drafting, to final polish. The key is treating it as a thinking partner with a strong brief, not a content vending machine. Here is my exact workflow, including the prompts I actually use.

Why I switched to Claude from other AI writing tools

I have used virtually every AI writing tool since GPT-3 in 2020. For pure writing quality, sustained context across long projects, and the ability to reason about tone and nuance, Claude 3.5 is the best I have used. It will tell you when your brief is unclear. It pushes back on weak arguments. I switched from a workflow split between ChatGPT and Jasper to Claude-primary about 9 months ago and have not looked back.

My complete content creation workflow with Claude

Stage 1 — Topic and angle research

Before writing anything, I use Claude to identify underserved angles. I feed it the topic and ask:

"I want to write about [topic] for an audience of [description]. Most existing content covers [X, Y, Z]. What angles are underserved? What questions is this audience asking that most posts fail to answer?"

Stage 2 — Brief development

I never go straight to drafting. I build a brief first: primary keyword, 3–5 secondary keywords, specific search intent, the direct answer I am promising the reader, H2 structure, and personal experience angles to include (crucial for E-E-A-T).

Stage 3 — First draft, section by section

Writing section by section gives more control and produces better output than requesting a full article in one go. For each section: "Write the [H2] section. ~[X] words. Voice: direct, experienced professional. Include: [specific point]. Do not use: [phrases to avoid]."

Stage 4 — Personal experience injection

This is the step most people skip. After Claude produces a section, I replace generic statements with specific, personal, verifiable experiences. Claude cannot fake lived experience — that is the competitive advantage. Example swap:

  • Generic (AI): "Many designers find that AI tools improve their workflow."
  • Specific (mine): "When I used Firefly generative fill on a licensed Marvel brief last year, what used to be a 3-hour background redraw took 20 minutes."

Stage 5 — FAQ generation

Every post ends with a FAQ section structured with FAQPage schema — pulled directly into Google AI Overviews and Perplexity answers. Prompt: "Based on this article, what are the 4 most common questions someone would still have after reading it? Write each question and a concise, direct 2–4 sentence answer."

Prompts that do not work (what I stopped using)

  • "Write me a blog post about X" — too vague, produces generic output every time.
  • "Make this more engaging" — Claude needs specifics: "Make the opening a direct challenge to a common assumption my audience holds about X."
  • "Write in a casual tone" — better: "Write as an experienced professional talking to a peer — confident, direct, occasional dry humour, no jargon."

Frequently asked questions

Is Claude free to use for content creation?

Claude has a free tier with limited usage. For professional content creation — where you need long context windows and sustained daily use — Claude Pro at £18/month is essential. The free tier hits rate limits quickly during a working session.

How do I stop Claude from sounding like AI?

Inject specific, verifiable, first-person details that only you could know. Also: "Do not use filler phrases like 'in today's world', 'it is worth noting', or 'in conclusion'." Claude responds well to specific constraints.

Can Claude replace a copywriter?

For volume content — product descriptions, email sequences — Claude significantly reduces the need for junior copywriting resource. For thought leadership requiring genuine expertise and lived experience, Claude is an accelerator for an experienced writer, not a replacement.