---
title: "Technical SEO in an AI website workflow"
description: "What a build workflow automatically gets right in technical SEO, and where editorial review begins. Without ranking promises."
source_name: "Gen8"
source_url: "https://gen8.app/en/kennis/technische-seo-in-een-ai-workflow/"
site_home: "https://gen8.app/"
source_description: "Managed AI service for SMEs. We set up, connect and manage; you stay in control."
locale: "en"
authors:
  - name: "Olivier Hokke"
    url: "https://gen8.app/en/kennis/author/olivier-hokke/"
copyright: "© 2026 Gen8 VOF · Dutch Chamber of Commerce 92978789 · Delft"
published_at: "2026-09-09T00:00:00.000Z"
modified_at: "2026-10-21T20:17:13.490Z"
---

# Technical SEO in an AI website workflow

Technical SEO has two types of work. One is mechanical and should never be manual. The other judges content and actually deteriorates when automated.

By Olivier Hokke · September 9, 2026 · Website

## The mechanical part

Part of technical SEO is simply consistently deriving something from what already exists. That belongs in the build, not a checklist. Every Gen8 site publication generates:

- `sitemap.xml` containing only published pages without `noindex`;
- `robots.txt`, allowing AI crawlers;
- `llms.txt` and an RSS feed;
- canonical links per page and language alternatives \(hreflang\) between translations;
- structured data: `BlogPosting` for article collections, `TechArticle` for documentation, `Article` for the rest.

One condition is yours to meet: that markup requires a title and publication date. An article without a date falls outside structured data, unnoticed until you miss it.

## What goes wrong automatically if left unchecked

Three things are technically automatable yet a bad idea to generate without supervision:

- **Bulk titles and meta descriptions.** A model produces polished, interchangeable sentences. Fifty pages with polished interchangeable sentences compete with each other and collectively say nothing. Generate a proposal and let a person decide what makes each page unique.
- **Internal links.** A generated link to a nonexistent page breaks on publication. A generated link to an existing but irrelevant page breaks nothing and helps nobody. Internal links deserve the same attention as their surrounding paragraph.
- **Alt text.** Describing what an image shows is automatable; why the image is there is not. Never generate alt text from a filename, and leave it empty for purely decorative images.

## The boundary

A useful summary: what follows from structure may be automatic; what concerns meaning receives editorial judgment. A sitemap follows from structure. A title concerns meaning. Hreflang follows from structure. Choosing which page gets `noindex` concerns meaning — and is also the setting most often left on accidentally.

## What this does not solve

A clean technical foundation makes you discoverable. Discoverable is not the same as ranking well. No generation step guarantees a position; that depends on how specific your content is and links other sites give you. Treat technical SEO as the whole project and in six months you have a technically impeccable site nobody visits.

## What to check periodically

Three questions, a few times yearly: does the sitemap include something that should not be there or omit something that should; do new pages have their own titles and descriptions rather than variations on the previous one; and is `noindex` still correct on pages carrying it? Fifteen minutes catches most issues.

## What to do with this

Open your site's sitemap and compare it with the page list you expect. Every difference — in either direction — is a concrete task, almost always a status setting or a leftover `noindex`. What the platform generates itself is described under [websites by agents](<https://gen8.app/en/platform/websites/>); the trade-off between AI building and building yourself is in [a website built with AI or by yourself](<https://gen8.app/en/kennis/website-met-ai-of-zelf-bouwen/>).

## Frequently asked questions

Must I enter a publication date for every article?

Yes. Article structured data requires a title and publication date; without a date, the page is excluded from that markup. Backdating is fine when transferring existing content.

Can I schedule pages for a later date?

No. A page with “published” status enters the next publication regardless of its date. Keep it as a draft and publish on the day itself.

## Want to apply this to your own process?

Bring one process to a conversation. We assess suitability, risk and the cost of the first step.

[Book an introduction](<https://gen8.app/en/contact/>)

- Reply within one business day
- No obligation