Automating WordPress Post Drafts with Make com and OpenAI
At a glance
- In shortA guide to generating clean HTML WordPress drafts automatically using Make and OpenAI.
- Best forSolo creators and bloggers running multiple WordPress sites
- Tools coveredMake.com, OpenAI, WordPress
- PricingMixed
- Skill levelIntermediate
- Key pointsAutomate repetitive administrative draft work while maintaining human review. · Force OpenAI to output raw HTML tags to prevent WordPress formatting errors. · Use WordPress Application Passwords for secure REST API authentication.
- Heads-upSecurity plugins may block REST API calls, throwing 401 or 403 Forbidden errors.
Last November, my Seoul apartment looked like an electronics recycling bin. I had four monitors glowing in the dark, three different AI text generators open in Chrome, and five draft tabs open in WordPress. I was manually copying generated text from ChatGPT, fixing headers, pasting excerpt summaries, uploading images, and setting tags across three niche content sites. It took me nearly 45 minutes per article just to perform administrative copy-pasting for content that was already written.
That week, I decided to fix the pipeline using Make (formerly Integromat) and OpenAI’s API. My goal was simple: input a topic into a spreadsheet, let AI draft the content formatted as clean HTML, and automatically create a styled draft in WordPress waiting for my final review.
If you run multiple sites or publish frequently as a solo creator, setting up a solid make com wordpress automation flow will save you hours of manual site management every week. Here is exactly how I built it, where I messed up, and how you can avoid the formatting nightmares I ran into.
Why Draft Automations Matter (and What They Cannot Do)
Before diving into the scenario setup, let us clear up a huge misconception about AI blogging. Fully automated, hands-off publishing directly to live WordPress sites is a quick way to destroy your search reputation and alienate readers. AI models hallucinate links, make up statistics, and occasionally repeat paragraphs.
My strategy relies on draft automation, not direct publishing. The automated pipeline takes care of repetitive administrative work:
- Formatting section headers with correct standard tags (<h2>, <h3>)
- Writing metadata, excerpts, and suggested tags
- Creating the post record directly inside the WordPress Gutenberg backend
- Setting the correct post category and author profile
Once the automation runs, I log into WordPress, read through the draft, fact-check statements, add real screenshots or custom media, insert affiliate links, and hit publish manually. This combines AI efficiency with human quality control.

Step-by-Step Setup: Building the Scenario in Make
To run this pipeline, you will need a Make account, an OpenAI API key, and a WordPress site with REST API access enabled (which is standard on almost all modern WordPress installs).
Step 1: Define the Trigger (Google Sheets or Airtable)
I start my scenarios with a simple Google Sheet or Notion database acting as my content calendar. For this example, let us use Google Sheets.
Create a spreadsheet with the following columns:
- Topic/Title Idea (e.g., “Best Portable Monitors for Nomads”)
- Target Keyword (e.g., “portable monitors seoul nomad”)
- Status (e.g., “Ready”, “In Progress”, “Done”)
In Make, add the Google Sheets: Watch Rows module. Set it to monitor your sheet and trigger whenever a row’s status column equals “Ready”. This ensures the automation only runs when you explicitly queue up a topic.
Step 2: Connect OpenAI to Generate Structured HTML Output
Next, add the OpenAI (ChatGPT): Create a Completion module. Select a model like GPT-4o or GPT-4o-mini depending on your speed and budget needs.
The critical mistake most people make here is letting OpenAI return plain markdown. WordPress handles raw markdown inconsistently through its REST API, often turning formatting into giant, unstyled text blocks. You must force OpenAI to output valid HTML tags.
Use a system prompt structured like this:
“You are an expert technical blog writer. Write a comprehensive article on the requested topic. You must output raw HTML code using only <p>, <h2>, <h3>, <ul>, and <ol> tags. Do not wrap the output in markdown code blocks like “`html. Output clean HTML text only.”
For the user prompt, pass the Topic and Target Keyword from your Google Sheets module.
Step 3: Create the WordPress Post Module
Add the WordPress: Create a Post module to your scenario. You will need to authenticate Make with your WordPress site. I strongly recommend using a WordPress Application Password rather than your primary admin login. You can generate one in your WordPress dashboard under Users > Profile > Application Passwords.
Once connected, map the fields as follows:
- Title: Map to the OpenAI output title or your Google Sheet topic.
- Content: Map directly to the HTML response text from your OpenAI module.
- Status: Set strictly to Draft.
- Format: Standard.
When Make runs this scenario, it makes a REST API call to your WordPress site and creates a clean draft ready for editing.
Comparing Automation Platforms for WordPress
You can achieve similar results using Zapier or n8n. Here is how Make compares based on my real-world usage across my sites:
| Platform | Ease of Setup | Cost Efficiency |
|---|---|---|
| Make.com | Moderate (Visual canvas) | High (Operation-based pricing) |
| Zapier | Very Easy | Lower (Task-based gets expensive fast) |
| n8n | Advanced (Self-hosted options) | Very High (If self-hosted) |
For solo creators, Make strikes the best balance between visual workflow debugging and affordable operations. Zapier quickly gets expensive when handling multi-step AI prompts, while n8n requires server maintenance that takes time away from publishing content.
Common Pitfalls and How I Fixed Them
When I first deployed this setup, I hit three major roadblocks that broke my site formatting or wasted API credits. Here is how to fix them:
1. The HTML Code Block Glitch
OpenAI loves wrapping code in markdown blocks (“`html … “`). If this gets sent directly to WordPress, your entire post will display inside a code preview box on your website frontend. To prevent this, add a text parser module in Make or explicitly tell OpenAI in the system prompt: “Do not include markdown ticks or backticks in your output.”
2. Missing Excerpts and Metadata
If you don’t generate an excerpt, WordPress grabs the first few sentences of your draft, which often looks messy in search previews or homepage archives. I solved this by adding a second prompt or using JSON response formats in OpenAI to request two distinct fields: “article_body” and “meta_description”. You can then map the meta description directly to the WordPress Excerpt field in Make.
3. REST API Authentication Failures
Security plugins like Wordfence or iThemes Security often block Make’s REST API calls by default. If your scenario throws a 401 or 403 Forbidden error, go into your security plugin settings and whitelist REST API access for authenticated application users.
Tool Pricing and Reality Check
Running this automation stack requires two main paid components outside of your web hosting:
- Make.com: Offers a free tier with around 1,000 operations per month, which is enough to test your scenarios. Paid plans start in the $9 to $10/month range for higher operation limits. Check the official Make pricing page for current tier structures.
- OpenAI API: Pay-as-you-go based on token usage. Generating a detailed 1,500-word draft with GPT-4o-mini costs fractions of a cent, while GPT-4o costs a few cents per run. Keep an eye on your usage limits in the OpenAI developer dashboard.

My Take
Automating WordPress drafts using Make and OpenAI completely eliminated the administrative fatigue that used to keep me stuck at my desk until midnight. By turning topic research into pre-formatted drafts automatically, I cut my production time per article by more than half.
However, automation is an assistant, not an author. If you rely entirely on Make to auto-publish raw AI text, your content will eventually feel bland, generic, and unhelpful. Use make com wordpress automation to eliminate copy-pasting, heading formatting, and manual post creation—then step in with your personal experience, screenshots, and unique perspective to make the content actually worth reading.
FAQ
Is it safe to publish AI drafts automatically without reviewing them?
No. Auto-publishing unedited AI drafts can hurt your brand and search rankings. AI models occasionally generate incorrect information, broken links, or generic phrasing. Always set your automation module to create drafts so you can review, edit, and fact-check before publishing.
Do I need a plugin to connect Make to WordPress?
No extra plugin is required if your site supports standard WordPress Application Passwords. You simply generate an Application Password inside your user profile in WordPress and use it to authenticate Make via the WordPress REST API.
Can Make automatically upload featured images to WordPress?
Yes. You can add a module like DALL-E or Midjourney (via API wrappers) in your Make scenario, generate an image, send the media file to the WordPress Upload Media module, and map the returned Image ID to the Featured Image field in your WordPress Post module.
More on WordPress & Blog SEO
- Image SEO Guide for SolopreneursExplains how to optimize alt text and filenames using AI tools to improve SEO and workflow.
- WordPress + AI Auto-Publishing SetupsExplores three levels of AI auto-publishing setups on WordPress, from simple drafts to automated pipelines.
- WordPress Cookie Consent Guide for SolopreneursA guide to understanding cookie consent laws and setting up compliant WordPress banners.
See every WordPress & Blog SEO post in our Guides & Tutorials guide.
Handy toolsAI tool ROI calculator · AI Tool Finder
