⛔ The Last Journey ⛔ 【AI Takeover】 The Quest of Building a Fully Autonomous Site Generator ❤️ Passive Income on Autopilot — Let's GO! ❤️

Status
Not open for further replies.
Working on my own system currently. Almost finished with the first step, namely automated domain registration and configuration (WHOIS privacy, nameservers, DNS records, autorenewal...). The APIs of NameCheap and Cloudflare are great. Probably due to their well written documentations as training data, ChatGPT immediately gives you the right GET/POST requests for specific tasks, e.g., "Set the SSL/TLS encryption mode of a domain in Cloudflare to 'Full' using the Cloudflare API in Python". Very handy.
 
Do you do SEO optimization for them?
Yes, I will do SEO, I am trying to create python script for this but right now I am confused on domain and how much articles go live per day. Also keyword scrapping is confusing for me.
 
As a programmer it was really interessting to read. All the best to you
 
Working on my own system currently. Almost finished with the first step, namely automated domain registration and configuration (WHOIS privacy, nameservers, DNS records, autorenewal...). The APIs of NameCheap and Cloudflare are great. Probably due to their well written documentations as training data, ChatGPT immediately gives you the right GET/POST requests for specific tasks, e.g., "Set the SSL/TLS encryption mode of a domain in Cloudflare to 'Full' using the Cloudflare API in Python". Very handy.
interesting stuff. i wonder if i can ask chatgpt to help me fix scripts when needed
 
yeah I'm not doing translations. Straight into 33 languages.
Do you first do the prompting and generating the text in English, and then asking the LLM to translate it in the 33 languages? Or do you just create the text 33 times in different languages (and thus might have different contents)
 
For multilingual sites are you using simply static or just plain WordPress with multilingual plugin? For multilingual site what you recommend, sub directory or sub domain structure ?
the LLM is generating straight into non-English content. I'm not using any plugins
 
Do you first do the prompting and generating the text in English, and then asking the LLM to translate it in the 33 languages? Or do you just create the text 33 times in different languages (and thus might have different contents)
we're not generating 33 versions. 1 site is 1 language. Straight to that language.
 
How does your script workflow look like when it comes to giving context to GPT to limit fake info generation?

For example do you scrape serps, extract <p> tag content, summarize the whole page and add it to the summary of the next and so on ?

Then include the merged summaries in the prompt let GPT take whatever information it needs from it?
 
Are you using multiple APIs to add relevant rich content to the blog post ? I would love some feedback on this too. I cannot figure out how to pull this off.

I might be completely off here but I think there might be some confusion on the definition of "rich content" and the term can be quite broad.
If you only create posts with texts, headers and a couple of images, everything becomes quite repetitive and it might be easier to spot that the site is in fact an AI site.

Figure out how to add bullet points, tables, forms, and other "types" of content whenever applicable and appropriate.
You can few-shot examples or fine-tune the LLM to figure out when a table might be good choice for particular "content section" or when to use bullet points...

This particular point where you have the oppurtunity to really differentiate your posts from everyone else is one area which I find very interesting.
 
I've made some improvements to the workflow for content generation.

Figured out a way to make GPT produce much longer form content.

Also, added support for 33 languages:

Code:
English
Spanish
French
Russian
German
Japanese
Italian
Dutch
Korean
Portuguese
Polish
Romanian
Greek
Swedish
Hungarian
Czech
Slovak
Danish
Finnish
Norwegian
Turkish
Hebrew
Croatian
Serbian
Slovenian
Estonian
Latvian
Mandarin Chinese
Lithuanian
Icelandic
Maltese
Bulgarian
Arabic

Here are 2 samples of content straight from the language model without any human editing:


Final content includes images, youtube videos, internal/external links, and rich content via embeddings. Still, I wanted to show you how it looks straight from the content generator without additional manipulation. Adding rich content is kinda easy.

Super excited about the progress. I never thought we would get this far. How will the Internet look in 5 years?
Hello. I want to ask something about something. How do you get traffic from Google? Do you know anything about SEO? How do you provide traffic on your sites that you create with artificial intelligence? I will be glad if you answer. Thanks.
 
hello J. @Sartre,

does it make sense to first generate junk into bh sites for rank tests -> churn to see SERPs -> rewrite quality ones to wh site -> use bh sites to link back wh site -> repeat.

hope it is explained clearly

Cheers,
Commoner
 
If you read my journey, I'm explaining that to create an article, we're using a chain of prompts and embeddings. There's also a complex NLP algorithm which cleans everything up and decides if we need to inject more info like box explanations, images, links, etc

adding rich content is just a lot of python scripting, natural language programming

the 'continue' thing is baked into the API. You just have to pay attention to the finish_reason variable from the API response.

@Sartre can you please give us general overview of your approach on how to insert rich content? Im creating long content made of headings and paragraphs, i would love to have some tables or lists in the mix.

Thank you in advance!
 
Oh, I also forgot to add another thing I do for contextual links.

Remember this isn't a "do this and you succeed, don't do it and you're flagged". It's very very likely now machine learning looking at everything. It's very likely there's very few set ranking factors these days, and the constant updates are re-trains of the machine learning which would make sense. The amount of updates are INSANE, and they didn't used to do that many. They used to be every 2-3 years, then yearly, then twice a year, now they're every 2-3 months.

The extra thing to do with all your contextual links is to avoid these spammy patterns you'll often see the most with cheap guest posts and pbns. "Why you should choose a plumber in NYC", then it contains a single link, even worse with an anchor "plumber nyc" lol.

Choose a title that's NOT directly related to the page you're linking to. This is actually not common with real links. Why would someone creating a page about forks, link to another page about forks? Sometimes yes, but now multiply this and all your contextuals are fork pages, linking to your fork page. And then people wonder why they got smashed with the Dec 14 spambrain update.

If your page is about forks, your article should be about a related topic. Dinner table etiquette, kitchen topics, food topics, cooking topics.

It also depends on the topic. If you're linking to a persian cat training page, you can have an article about persian cats, but don't have an article about persian cat training. Or, have an article about training cats in general. Always something a little off to the left.

Add internal links to your backlink. Harder for guest posts, but you can ask.

Add another 2-3 external links to the article. You will NOT dilute the power. This is not 2010 ;-)

Add 1-2 internal link FROM the article. Ie, link to a couple of other internal pages. This, again, is a sign of a real article.

An article that's not real will have 1-2 links, no internal links to it, no internal links from it.
Thanks, my bot currently searches related links from google scholar then adds the post like this

References:
[Google Scholar Link 1]
[Google Scholar Link 2]

I also already created a wordpress plugin that auto inserts anchor link based on old post using nlp in php. So, I don't have any problems with internal links.

My only problem is external links, should I google the keyword and get the link of the top result on google? or stay from google scholar?

Also, how should I place the external link? My wordpress plugin inserts anchor link to text, whilst the external links are posted directly at the bottom of the page like this:
References:
[Google Scholar Link 1]
[Google Scholar Link 2]

Sorry I am new to SEO but I am not new in programming. Thanks
 
you're getting a captcha scraping amazon? Amazon Advertising API has pretty good limits, but you need sales to get approved. I used to scrape 4 million products daily for a deals site.
May I know if you are inserting your amazon affiliate links automatically to your blogs? Or you are focused on informational user itent rather than transactional or commercial?
Up to 250k tokens generated every minute
Davinci costs $0.0200 / 1K tokens, does that mean you are spending 5000 US$ per minute?

after 1-2 months evaluate if Google is picking it up and remove "bad" content
How do you know if your content does not have any clicks nor impression or it's a bad article? May I know what API are you using?

Also, may I know what Author are you placing on your site to build E-EAT? Are you using one author for all sites? Are these authors fake and you just made it up? Or your name is the author with your profile picture?

Also, May I know the Design of your website? Does your 1k sites use the same themes?

Thanks
 
@Sartre @splishsplash May I also know if you guys are using the information on the internet such as the serps featured snippet and meta data?
Do you include these on the AI prompts? Because sometimes AI returns incorrect facts, getting metadata and featured snippets on google might be helpful.
Thanks
 
Thanks, my bot currently searches related links from google scholar then adds the post like this

References:
[Google Scholar Link 1]
[Google Scholar Link 2]

I also already created a wordpress plugin that auto inserts anchor link based on old post using nlp in php. So, I don't have any problems with internal links.

My only problem is external links, should I google the keyword and get the link of the top result on google? or stay from google scholar?

Also, how should I place the external link? My wordpress plugin inserts anchor link to text, whilst the external links are posted directly at the bottom of the page like this:
References:
[Google Scholar Link 1]
[Google Scholar Link 2]

Sorry I am new to SEO but I am not new in programming. Thanks
> Google you keyword or summary of the article.
> extract the snippets of top 10 results.
> use embeddings to compare similarity of each snippet to your keyword/summary.
> link the one with highest similarity score.

This is my approach, just works fine for me.

Interested in understanding the workflow of your internal linking plugin.
 
> Google you keyword or summary of the article.
> extract the snippets of top 10 results.
> use embeddings to compare similarity of each snippet to your keyword/summary.
> link the one with highest similarity score.

This is my approach, just works fine for me.

Interested in understanding the workflow of your internal linking plugin.
Thanks, are you using anchor links to your texts? Or you have dedicated references at the bottom of your page and you link directly in there?

When a new post was published, my wordpress plugin works in background to scan all title of old posts then compare the titles of old posts to each phrases of my paragraph, if relevancy was detected, then link to that phrase
 
Thanks, are you using anchor links to your texts? Or you have dedicated references at the bottom of your page and you link directly in there?

When a new post was published, my wordpress plugin works in background to scan all title of old posts then compare the titles of old posts to each phrases of my paragraph, if relevancy was detected, then link to that phrase
I am still experimenting, but I think I will be using anchor links.

How much time it takes for your plugin to complete interlinking of one article (let's say article is 2000 words long)
 
Status
Not open for further replies.
Back
Top