How I work with keys

1. I feed the collected raw keys to the promptus.
2. Intents are defined on the basis of keys
3. entities are defined on the basis of intents
4. The list of found entities is extended and supplemented with related and implicit entities
5. Entities are grouped into domains
6. New intents are defined for all domains and all entities
7. Intents are clustered
8. A taxonomy based on entities and intents is constructed
9. The taxonomy is filled with content (generachim according to mega rules - see prompto-editing)
10. Embed some of the keys into the content (if they are not already there)
11. We're doing a self-check on all this stuff.
12. Introduce linking in the content
13. We're doing a self-check of all this stuff
14. Uploading to the site, proofreading

That's just briefly. If a little more detailed - watch and read season 1 and 2 of Promptology - a lot of this is covered there.

This is what I wrote in general. Here my friend says that LLM equalized the quality of content for the entire Internet and everyone has the same average.

No way.

I gave my combat prompts to my comrades, recorded a video of how to work with them, how to adjust them if necessary. I showed all the settings in LLM.

And the quality of the output content is completely different between me and my comrades. Even on my combat prompts. I thought about it and realized it's two things:
1. Input data to the prompt
2. Proofreading (to a lesser extent).

You could write a whole book about how to prepare data for prompt. The way I divide it is my personal know-how, which I have been using successfully for more than a year. It can't be described in the articles in the channel, and it's not worth it.

In general, use the scheme, improve it. Read Promptology.