You type “write a blog post about choosing web hosting” into ChatGPT and get 800 smooth, confident words back in ten seconds. Then you read it. “In today’s digital landscape…” Every point is technically true. None of it is useful. It could have been published on any of a thousand sites, and that’s the problem.
AI tools can save you hours, but only if you use them for the right parts of the job. Here’s a workflow that does that.
Why AI Drafts Come Out Generic
Language models are trained to predict likely text. Ask a broad question and you get the most average answer, a blend of everything already written on the topic. Readers and search engines have seen that average answer thousands of times already.
What AI doesn’t have is:
- Your experience: what actually happened when you tried something
- Your opinions: which option you’d pick and why
- Specific, current details: your test results, screenshots, prices you checked today
- Your readers: the questions your audience keeps asking
Those four things are what make a post worth reading. Your workflow should keep them in human hands.
Where AI Helps and Where It Hurts
| Task | Use AI? | Why |
|---|---|---|
| Brainstorming angles and subtopics | Yes | Fast, broad coverage of ideas |
| Building an outline | Yes, then edit it | Good skeleton, needs your judgement |
| Writing your intro and opinions | No | This is where your voice lives |
| Drafting explanations of standard concepts | Carefully | Useful, but check every fact |
| Rewording clunky sentences | Yes | A good editing assistant |
| Facts, prices, statistics | Never trust blindly | Can be outdated or simply made up |
| Titles and meta descriptions | Yes, as options | Pick and adjust the best one |
The 6-Step Workflow
Step 1: Start with your notes, not a prompt
Before opening any AI tool, spend 10 minutes writing bullet points: what you know, what you’ve tried, what surprised you, what you’d recommend. Rough is fine. These notes are the part no AI can produce.
Step 2: Ask for an outline, with context
Give the tool your audience, your goal and your notes. For example:
“I’m writing for beginners starting their first WordPress blog. The post answers: ‘Which type of hosting should I choose?’ Here are my notes: [paste]. Suggest an H2/H3 outline that covers what a beginner needs to decide, in a logical order. Flag any important subtopic my notes miss.”
The last line is where AI really earns its keep: spotting gaps.
Step 3: Edit the outline yourself
Delete sections that are filler. Reorder things to match how a real reader thinks. Add sections for your examples and opinions. This takes five minutes and shapes the whole post.
Step 4: Write the key sections yourself
Write the intro, your recommendations, examples and verdicts in your own words. Use AI for the standard explanatory sections if you like, one section at a time, with specific instructions:
“Explain what a VPS is in under 120 words for a complete beginner. Use a simple analogy. No hype words.”
Short, specific prompts give you far better results than “write the whole post”.
Step 5: Fact-check everything
Check every number, price, feature claim and statistic against the original source. AI tools can confidently state things that are outdated or completely made up. If you can’t verify a claim, cut it.
Step 6: Edit for voice
Read the post aloud. Cut anything you wouldn’t say to a friend. Watch for common AI tells:
- “In today’s digital world” / “In the ever-evolving landscape”
- “Dive in”, “delve”, “unlock”, “elevate”, “game-changer”
- Every section ending with a neat one-line summary
- Lots of confident claims with no specifics behind them
Example: Generic vs Edited
Generic AI output: “Choosing the right hosting provider is crucial for your website’s success. There are many factors to consider, including speed, reliability, and support.”
Edited with experience: “I switched hosts twice in my first year. The first time because of a surprise renewal bill, the second because the site kept timing out every afternoon. Both problems were visible before I signed up, if I’d known where to look.”
The second version gives the reader a reason to keep going.
Build a Reusable Prompt Library
Once you find prompts that work, save them so you don’t have to reinvent them every time. A small library of five or six is enough:
- Outline prompt: your audience, the question, your notes, and a request to flag gaps
- Explainer prompt: “Explain [concept] for a complete beginner in under [X] words, with one analogy, no hype words”
- Editing prompt: “Tighten this section. Keep my wording where possible, cut repetition, and point out anything unclear”
- Title prompt: “Give me 10 title options for this post. Specific, under 60 characters, no clickbait”
- Meta description prompt: “Write 3 meta descriptions under 150 characters that state the benefit clearly”
Add a short description of your site’s tone to each prompt, such as “practical, friendly, no jargon, honest about downsides”. This one line makes a noticeable difference to the output.
Is This Allowed by Google?
Google’s guidance focuses on whether content is helpful and made for people, not on whether AI was involved. It does act against mass-produced, low-value content made mainly to game rankings. We cover this in detail in will Google penalise AI-written content?
Takeaway
Let AI handle the parts where speed matters and anyone’s answer would do, like outlines, rewording and gap-spotting. Keep the parts that make a post yours (experience, opinions and verified facts) for yourself. For more tools that fit into this workflow, see AI tools that save bloggers real time.