How to use AI to write a blog post (that isn’t slop)
🔎 Nguồn: Chuyên ngành · Bài viết chia sẻ chuyên môn khách quan, có dẫn chứng số liệu khảo sát.
Original (en)
Outline your company's marketing strategy in one simple, coherent plan.

Like a hammer or even a hair dryer, AI is just a tool. It needs to be wielded skillfully by an expert to achieve greatness. And in a world increasingly full of “AI slop,” this is especially true when exploring how to use AI to write a blog post.
When using AI to write blog posts, you can’t just sit back and click copy-paste and publish. You need to create a thorough brief, build an outline, draft one section at a time, and then edit with the emotion, lived experience, and discernment only a human can.
AI can produce quantity, no doubt, but its quality is up for debate — at least without your guidance. AI is wonderful for structure and first drafts, but only you can bring the expertise, voice, and final quality assurance that produces great content. And this guide will show you how.
You’ll get a five-step process for creating blog posts with AI, human-review checkpoints, prompt templates, and tips for keeping your voice and avoiding the “slop” label .
%20hubspot%20ai%20content%20writer%20editing%20email%20copy%201%20(1).webp?)
Ideate, create, and share content. Easily create...
Generative AI is powerful and useful. It can analyze information and produce materials faster than any human could by hand, which is likely why 80% of marketers now use AI for content creation in some form.
AI slop is low-quality, mass-produced digital content created using generative AI with little to no human input or creative thought. (I’m sure you’ve seen the fliers .)
View this post on Instagram A post shared by @sandsvisual
Users and businesses sharing it have been ridiculed as low-effort, unoriginal, or even “fake.” Harsh, yes, but also easy to avoid.
Listen: It’s easy to get caught up in the speed and “polished” finished products that AI generates in seconds; I don’t blame anyone. But at this point, we, as marketers and consumers, have seen enough of its generic output to know we need to do better.
AI is a collaborator, not a replacement for your judgment, creativity, experience, or effort.
Now, I know what many of you are thinking: “Why would I do all that work when AI can do it for me?” And I get it. You want to save time and labor, but to quote former Director of the HubSpot Blog Karla Hesterberg : “The only truly effortless [AI] process produces slop.”
I’m not saying don’t use AI. I’m saying use your skill with it.
This is something we learned firsthand by adopting AI into our editorial process at HubSpot , and it heavily influenced the steps in this guide. Tools like HubSpot’s brand voice and AI content writer were built with this all in mind.
AI is like an eager new intern. It’s fast and takes direction, but it doesn’t have your lived expertise or knowledge of your customer, data, and standards. The steps we share in this guide are designed to compensate for that and ensure you maintain your brand voice.
Each step has something you give AI, a choice you make, something you get back, and a moment when a human steps in, based on my decade as a content editor and strategist.
You can follow these steps in common tools like Claude or ChatGPT, or in HubSpot, which offers multiple options.
HubSpot’s free AI Content Writer can draft blog posts, outlines, and more inside the tools you already use. For a full blog workflow, Content Agent builds blog posts, social content, and landing pages in your brand voice. It lives in Agent Hub , which is included in Starter, Professional, and Enterprise editions.
Now that you know what you need, let’s break down our five steps.
Before you turn to an AI content creation tool, you need the right input. Think of these as the creative brief you send an intern.
If you use HubSpot, Agent Hub runs on your CRM data. HubSpot says its third-party AI providers are contractually barred from training on your data. You also control what AI can access. You can read how it works in the HubSpot Trust Center .
Pro tip: Put all your consistent inputs (e.g., brand voice, company history) on a one-pager and save it as a template you can reuse for every post. This saves you time and maintains consistency across different editors.
Need a blog topic? Try HubSpot’s free Blog Idea Generator to spark ideas, then pick the one that fits your goal.
An outline is the heading and talking points plan that AI follows. Skip it, and drafts wander. Start with your brief of inputs. Then ask AI to turn it into an outline. Pair each heading with the reader’s question it answers and the source that supports it.
To get an outline you can actually use, get specific. The more ambiguity you leave, the more likely the finished product will miss the mark.
Look at the difference between a vague prompt and a specific prompt.
“Write an outline about email marketing.”
“Create three outlines for a 1,500-word post for small-business owners who are new to email. Primary keyword: email marketing for beginners. Each H2 must answer one question. Start with a direct answer.”
Try entering your topic in HubSpot’s free AI Content Writer and asking for an outline. Then edit the headings to match your brief.
Want more prompt ideas? Our guide to writing AI prompts shows how specific details lead to better results.
Once you have a solid outline, you can work with AI to generate the main sections of your article.
You wouldn’t just give a new intern a brief and say “go,” right? You’d likely have them complete the assignment in small chunks to make sure they’re on the right track. Approach A the same way.
Draft sections one by one to increase focus and maintain quality, or even better, draft select sections rather than all of them. Review each one in depth (more on that in the next section). Then move on to the next.
With challenges like plagiarism and inaccuracy, this chunked approach to writing a blog post with AI is worth the effort.
Personally, I prefer to use AI-generated copy simply as a very rough example that gets my creative juices flowing. I edit and revise to fit my own voice, adding my original thoughts, knowledge, and examples, among other things.
You can follow this drafting workflow in Content Hub , HubSpot’s content marketing software. The free AI Content Writer lets you enter a prompt or use slash commands to generate text on demand. Highlight a paragraph, and you can ask it to edit or rephrase the copy.
But remember: Drafts written in chunks still need human editing.
Ideate, create, and share content. Easily create...
Even your best human-written draft shouldn’t be published without thorough editing. Trust, quality, and originality are ensured in the edit, and that goes double for AI-assisted blog posts.
Before anything goes live, have a human team member check four key things:
AI has a way of sounding extra confident, even when it’s wrong.
Google says this is because generative models predict a likely sequence of words instead of retrieving facts. So, the output can contain errors, or what we call “hallucinations.” NIST, the U.S. standards agency, lists this risk (it calls it “confabulation”) among the 12 risks associated with generative AI .
For high-stakes topics (i.e., legal, medical, financial, or anything that names a specific person or company), add a second reviewer. People can slip up, too, and when it comes to risky subject matter, it’s wise to double-check.
Original means more than not copied. It means your content adds something new. AI-generated content, by its very nature, is derivative, so editing is a must to deliver fresh value.
Before you publish any AI-assisted content, run through these checks:
Also, eliminate common AI writing patterns by:
Including unexpected examples is one of my favorite ways to add originality and personality to my articles. I mean, come on: how often do you see Bollywood mentioned in marketing blogs?
For more edits that make AI drafts sound human, see our guide to humanizing AI content .
More than anything, make sure that your AI-assisted content is true and consistent with your brand voice and perspective.
Using language or sharing opinions that contradict what your brand has already established can cause confusion, and any confusion about your brand’s beliefs can lead to distrust.
To help maintain your brand when using AI to write a blog post, create a voice profile. A voice profile is a short description of how you write and what you sound like. Think of it as a style guide you include in every prompt.
Voice: Friendly, expert, plain-spoken.
Sentences: Under 20 words. Mix short and medium.
Explain industry jargon or acronyms the first time they appear.
Avoid: “game-changer,” “unlock,” “in today’s fast-paced world.”
HubSpot’s Brand Voice learns your style from your writing samples and generates this guide for you. After that, you can export it to share with AI tools or have it automatically apply your style in supported tools, such as Marketing Hub. You can also set up a brand voice using AI , then fine-tune its personality and tone. Features can vary by plan, so check what your subscription includes.
Need a deeper dive? See our guide on AI brand voice training .
Litmus is a great example of skimmable formatting in action.
AI is a handy assistant for small, repetitive SEO tasks. Ask it for options to edit for:
You’ve drafted, edited, and optimized for getting found. Now it’s time to publish. Check your final boxes in four main steps.
1. Do a final check. Read the whole post once, start to finish, as a reader would. Look for voice slips, vague claims, and broken links, and edit as needed. Then confirm the basics:
2. Decide on disclosure. Some teams add a short note on how AI was used in their writing process. Google says sharing how content was created can give readers helpful context and set expectations. Check your company policy and local rules first. (See the FAQ below.)
3. Publish & measure what matters. In HubSpot’s 2026 State of Marketing report , measuring ROI was the top challenge, cited by 33% of respondents. So, keep it simple. Pick three numbers that reflect your goals. A good baseline to consider:
HubSpot can show many of these in one place with its marketing analytics . To see how your brand shows up in AI answers, try the free AEO Grader or HubSpot AEO , currently in beta.
4. Learn from every post. Keep a short prompt log:
Turn those repeat edits into new rules in your voice profile. Your next draft will start closer to being done.
Ready to put this process into action ? Try HubSpot’s AI Content Writer to draft your next post, then follow the five steps above to make it yours.
Ideate, create, and share content. Easily create...
Below is a series of prompts that work with HubSpot’s AI Content Writer and most other AI tools. Replace everything in [brackets] before you send, but remember that whatever you get back is just a starting point.
Every prompt needs your audience, your goal, your approved sources, and your voice profile. Without them, you’ll get generic results. So, include this block before you start drafting anything:
Audience: [who they are and what they already know]
Goal: [what the reader should learn or do]
Approved sources: [paste links or notes. Use only these.]
Voice profile: [paste your voice profile]
Using the details above, create a one-page brief for a blog post about [topic]. Include the reader’s main question, the angle, a two-sentence direct answer, an outline with H2s and H3s, and the approved source that supports each section. Flag anything the sources don’t support.
Give me three outline options for a [word count] blog post about [topic]. Each H2 must answer one reader question. Put the direct answer first. Don’t add sections that the approved sources can’t support.
Write the section titled “[H2 heading]” in [150-250] words. Use short sentences. Use only the approved sources. Mark any claim that needs a source check with [CHECK].
Rewrite the paragraph below in my voice profile. Keep every fact the same. Don’t add new claims. Cut filler and cliché phrases.
List every factual claim in the draft below: stats, names, dates, quotes, and product features. For each one, tell me what kind of primary source would confirm it. Don’t confirm anything yourself.
Write five meta descriptions of around 150 characters for a post titled “[title]”. Use the primary keyword “[keyword]” once. Promise only what the post delivers.
Below are my draft and a list of related posts with URLs. Suggest up to five places to add internal links. For each, give the sentence, the post, and descriptive anchor text. Only suggest links that truly help the reader.
[paste related posts and URLs]
Use these prompts strategically and flesh them out with your own original thoughts. For more on writing prompts that work, see our guide on how to write AI prompts .
Pro tip: Save your best prompts in a shared doc. Add notes on what each one is good for. Your whole team will get better results, faster.
Google hasn’t explicitly said that it penalizes AI-written content, but its quality standards and spam policy on content abuse at scale target mass-produced pages that add little value for users, regardless of how they’re made. Using AI to generate many pages without adding value is one of its named examples. Publish helpful, accurate, edited posts, and you’re on the right side of that line. For more, see Google’s guidance on generative AI content .
Start with what readers would find useful. Google says sharing how a piece of content was created can give readers helpful context and establish expectations. For example, take “We used AI to help draft this post. Our editors reviewed and fact-checked it.”
Laws are also changing. The EU AI Act’s transparency rules under Article 50 took effect on August 2, 2026, and cover AI-generated text published on matters of public interest, with an exception for text that receives human review or editorial control and is attributed to a person or organization. Whether these rules apply to you depends on your content and audience. Ask your legal team.
All of it. But these parts need extra care:
Unfortunately, you can’t eliminate AI hallucinations, but you can reduce the risk. Give AI approved sources and tell it to use only those. It may not always listen, but do your due diligence. Draft in small sections. Ask it to mark any claim that needs checking. Then verify every fact yourself.
Hallucinations occur because generative models predict likely words rather than retrieve facts , as Google explains. NIST lists confabulation among the key risks of generative AI . Human review is the safeguard.
I can’t say it enough: AI is a tool. It’s a powerful tool, but a tool nevertheless. A hammer can’t build the house, and AI can’t write a great blog post on its own.
Use AI for monotonous things like general info, structuring, outlining, and small repetitive tasks. And save the work that matters most for your people: unique expertise, lived experience, and discernment. That’s what readers want more of, and it’s what AI can’t fake.
Ready to try it? Start drafting with HubSpot’s AI Content Writer , then make every word your own.
Ideate, create, and share content. Easily create...










