Old approach was to:
Grab a list of websites in same niche that are doing well scrap their content then spin that content (using google translate in my case) and post it.
There used to be a software back in the day called multiblogger that had more options like it took some keywords generated relevant lSI keywords, created content, and also posted the content.
It also had option to create content from a list of sitemap url's of different websites.
Content creation was using 3rd party spinning services like WordAI and it also had inbuild sytsem which was pretty good standard at that time.
For images royalty free images or scraped images were used.
Expired articles and their images were also a good source back then.
New approach:
As mentioned in my previous reply the new reply is semi-automated with the content planning being the most important part.
I use tools like seo powersuite, serpstat, and google search with seo extensions to create a topical relevant content plan with clustered topics/keywords.
This is done manually and is the most important part here.
Then content is generated with AI and optimized with AI after comparing the optimization with top serp result pages.
This process involves custom scripts and perfectly working prompts for the AI.
Once the optimization score for article is good it gets posted and I review myself and with a VA's on every Weekend.
For images I still rely on free CC and royalty free images.
Another most important change with my new approach is I focus more quality rather than quantity.
So, instead of thousands of articles posted per day it's now some hundreds posted per week with more relevance and better quality.
Plus based on data from seo tools non-performing articles are improved to meet the standards.
It's not an ideal white hat approach but a careful measured approach.