DeepSeek post writer

Hey, I have a question about my design flow.
My assumption is that the result of his work will be a post posted on one of my wordpress sites. Without a photo. Just html formatted content.
The first assumption is that I want to do a SERP scrape on a topic. I will get a list of pages regarding the given topic. I then want to send this to the language model - Deepseek. I am wondering about such a thing:

  1. do I send the data separately for each scraped page?
  2. Should I send the data from all the scraped pages at once?
    The idea is that deepseek then writes an article on a given topic based on the data provided and its training data.

I wonder if I send the data separately will deepseek remember that I sent it? Will each call to the api be a separate conversation, just like in deepseek chat?

How do you solve this for a workfow like this?

Second question for those using DeepSeek to write their articles along with SERP analysis: does your workflow write the article in steps? Meaning:

  1. step one - research from the data received from the serps
  2. based on the above data, determining the number and content of headings
  3. writing content for the first headline, bearing in mind the input data
  4. write the content for the second headline bearing in mind the input data and what has already been written.
    How do I design this to work? I have a very large number of posts to write and need to start constructing something. I can also go into collaboration with someone and do the flow together.