n8n blogging pipeline: review the source, draft and published result

Build the request-to-draft route

The 11 May 2025 tutorial combines n8n, OpenAI, SEOwriting.ai and WordPress for blog creation. It begins with a chatbot and later adds Brave Search and Airtable for research and preferences. These parts do different jobs: the request sets the topic, generation creates a draft, search supplies material to check and saved preferences guide later work.

1. Follow the chatbot workflow at 1:52 and identify the topic, audience and expected article output.
2. Use the retained generation resource: https://lukejb.short.gy/seowriting.ai.
3. At 3:33, inspect the generated article’s structure and factual claims before moving it to the publication stage.

Review WordPress delivery as a separate integration

WordPress work begins at 8:44. A draft in a generation tool and the final page on a site are distinct outputs. Formatting, images, links and publication state can change during transfer, so keep the version you reviewed identifiable when the workflow passes content to WordPress.

1. Compare the outgoing title and body with the reviewed draft and identify the intended WordPress destination.
2. Inspect the resulting page or draft for headings, paragraph order, image placement and working links.
3. Check the actual publication state rather than assuming a successful API response means the article is live or ready for readers.
4. Keep uncertain outcomes recorded before retrying to avoid creating duplicate posts.

Add research and reusable preferences without inventing learning

Brave Search and Airtable appear at 26:12, with the final workflow at 48:02. The description calls this self-learning and mentions keyword memory. Saving preferences or prior keywords is useful context management; it does not by itself establish that the underlying model has been retrained or that a new draft’s facts are correct.

1. Inspect the search material behind important claims and retain source dates where they affect the article.
2. Keep preferred style, previously used keywords and source facts in distinguishable records so one cannot silently replace another.
3. Run a second article request and check whether the intended preferences are actually reused without copying irrelevant details from the first draft.
4. Review content and page output again before expanding recurrence or volume.

Check the production assumptions

The source promotes free deployment, SEO optimisation and a 25% LUKE25 offer. These are historical claims and terms, not proof of current all-in cost or search performance. A functioning pipeline still needs accurate content and an audience-relevant reason for each article.

1. Track generation, search, hosting and publication costs separately for your actual setup.
2. Record what was reviewed, what remains unresolved and which version was delivered before calling the run complete.

The takeaway

The valuable automation preserves a checkable editorial chain; saved preferences and generated SEO fields do not replace research or review.