[Journey] AI generated content indexed by Google

1. Clean up the site (WordPress): optimize pages, fix broken links, and update outdated content.
2. Set up Google Search Console and GTM ect.
3. Set up automatic publishing using tools like Google Sheets and Make.

1. Set up the AI System. (see below..)
2. Generate 10 articles.
3. Review AI-generated articles.
4. Adjust the prompts and parameters.
5. Repeat steps 1-4 until satisfied.

1. Generate 10 articles and apply human corrections.
2. Publish on the web.
3. Fine-tune the custom GPT model with 10 generated and corrected articles.
4. Generate another 10 articles and apply human corrections.
5. Publish on the web.
6. Fine-tune the custom GPT model with the 20 revised articles.
7. Generate another 10 articles and apply human corrections.
8. Continue this process until the AI text no longer needs reviews and is indexed by Google.

1. Find an trending article related to a my niche.
2. Provide the link to Perplexity with the prompt: ‘Identify the key message and summarise the article in a three sentences.’
3. Send the article summary to Perplexity with the prompt: ‘Do an extensive research and prepare a comprehensive paper to support article writing.’
4. Send the research results to GPT with the prompt: ‘Prepare an article following the specified format and structure.’
5. Send selected paragraphs from the article to DALL-E with the prompt: ‘Create graphics to illustrate the ideas in the text.’

The actual prompts are much more complex, of course. The one I'm using for the research part alone is now over 500 words.
This means that cost optimisation and batch processing will also be parts of the project


That’s all for now. Stage 0 is almost done. I’ll write an update in a few days, after a few articles have been published.
In the meantime, if you have any questions or suggestions, let’s talk!

Daniel16 said:

Do you use any tools to convert AI content to Humanize them?

Click to expand...

blogmen said:

Just wanted to point out that reaching a certain text quality is not the reason why Google will decide to index your content.

Google doesn't index based on the text quality alone, it can index junk content just fine depending on the target keywords, search intent and authority.

Click to expand...

jackyok said:

For the “humanizing” part of the process responsible is the fine-tuning loop, where I feed the model with heavily rewritten texts. These texts will be rewritten by humans, so the model can learn the specific style and wording of a given business or person.

I can rewrite AI-generated texts quickly and effectively, so I won’t be using any additional tools. However, I’m designing this system as a service for my clients, which means they’ll need to do the rewriting themselves (to reflect their own communication style). In this scenario, the humanising tool will be used to spot and highlight AI leaks and red flags, just to help them in the process. To clarify: they will be working on the original texts, not on the version humanized by the tool.

In general, using only humanizing tools for the rewriting step in the learning loop is an idea worth considering, but only for projects where personal style isn’t essential or the scale is too large for manual rewriting.

I don’t plan to use any humanizing tools in the final model—the goal of this project is to train a model that can generate texts without needs of extra “humanization.” If it does, that means I’ve failed this journey

Keep in mind, that the goal is to be indexed by Google. Their official statement is that they don’t care whether content is created by AI or HI (Human Intelligence )— it just needs to be valuable to readers. So passing an AI test isn’t crucial—getting indexed by Google is the ultimate test



You’re right—some SEO stuff will be addressed in the prompts. There will be more about SEO in Stage 2. And of course, topics like link building and authority management are important, but they’re out the scope of these tests.

Click to expand...