How to run a blog and SEO engine with AI.

I use this to run meritt's content marketing - authoritative, SEO-optimised blog posts that bring in buyers and candidates, kept fresh by an enrichment loop. Two jobs: write a new post on a sales-hiring topic (B2B for buyers, B2C for candidates), and re-read existing posts to enrich them and audit their search performance. I reach for it when I want a new piece of content, and the enrichment loop runs on a schedule to keep the catalogue alive.

How I use this, and what for

I use this to run meritt's content marketing - authoritative, SEO-optimised blog posts that bring in buyers and candidates, kept fresh by an enrichment loop. Two jobs: write a new post on a sales-hiring topic (B2B for buyers, B2C for candidates), and re-read existing posts to enrich them and audit their search performance. I reach for it when I want a new piece of content, and the enrichment loop runs on a schedule to keep the catalogue alive.

What the AI actually does

You give it a topic or an insight question. The AI writes the full post - voice, persona, structure, FAQ, internal links, one original meritt stat or point of view, and the practical AI section - QAs it, makes the header, and publishes it. Then the loop re-reads old posts and enriches them. The manual version is a blank page per post and a catalogue that quietly goes stale. This is a finished, published post and a self-maintaining back catalogue.

The publish target is built for this. Payload CMS is the open-source, TypeScript-native CMS that installs directly into a Next.js app folder, so the CMS and the site are one project and one deploy - the blog-posts collection in the meritt-website repo is where posts live and go live. The header image comes from OpenAI image generation, kept clean and text-free so the site renders the title.

The tools and APIs you need

• Payload CMS
• Next.js
• OpenAI images

The right prompts

Write a new buyer post

```
/b2b-blog-creator
Write a post on "how to screen SDRs for coachability, not just confidence."
Buyer audience. Include the four-trait angle, an original meritt POV, an
FAQ, 3-5 internal links, and a practical "how to use AI to do this" section
with a paste-ready prompt.
```

Audit a live post

```
/seo-geo-aeo
Audit https://www.meritt.io/blog/[slug] for SEO, GEO and AEO.
Give me the fixes ranked by impact - meta, schema, snippet readiness,
and what to add for AI-search visibility.
```

How to do it

Pick the audience

Buyer topic goes to `b2b-blog-creator`. Candidate topic goes to `b2c-blog-creator`. Give it the topic or the insight question.

Let it write the full package:

voice, persona, four-trait framework where relevant, SEO structure, FAQ, internal links, one original meritt stat or POV, and the practical AI section.

QA it

with `blog-editor-qa` - tighten, fact-check, brand-check.

Make the header

with `blog-header-creator` (clean, text-free, on-brand).

Publish

with `blog-publish` to the [Payload](https://payloadcms.com/) `blog-posts` collection.

Audit and enrich

Run `seo-geo-aeo` on the live URL. Send researched alternatives to the knowledge base with `add-subject`, not into the post.

See the difference

Let AI carry the heavy lifting; you keep the judgement and the final say.

How you'll know it's working

• A post that ranks and reads well, in the right voice for its audience.
• Every post carries a practical AI section a reader can act on.
• The deep catalogue of alternatives lives in the skills database, not bloating the blog.
• Old posts get enriched on a loop instead of going stale.

Questions people ask