Curious to know what you're using in the backend? Did you do any finetuning? Happy to share my process (what I did was scrape top 1mill websites and use that as training data for names on T5 nlp model)
I am using OpenAI GPT3 with a single (but long) prompt, explaining how to come up with the names that I think would be interesting.
Honestly, people who just about "prompt engineering" are pretty on point. It requires a bit of experimentation, but in this case I didn't have to do any fine tuning of the model itself, just the prompt.
I might come up with an array of different prompts to alter results, depends on user preferences.
My numbers for affiliate earnings were relatively low, which is why im asking. OpenAI API is expensive.