How do you scale local SEO pages with AI without creating repetitive semantic patterns?

We’re running into a weird problem while scaling a location-based content site with hundreds of profile pages across different cities.

The issue isn’t keyword stuffing anymore — it’s semantic repetition.

Most AI tools (ChatGPT, Gemini, Claude, etc.) keep generating the same narrative structure over and over:

– “she spends time between…”
– “her style is connected to…”
– “luxury lifestyle”
– “discreet settings”
– “private dinners”
– etc.

Even when the cities and places change, the semantic DNA still feels identical.

We need pages that:
– rank locally for different cities,
– still include local entities and commercial intent,
– but DON’T feel programmatic or AI-generated.

The hardest part is balancing:
– local SEO relevance,
– editorial/brand tone,
– semantic uniqueness,
– and scalable production.

Has anyone found:
– a better AI workflow,
– prompting method,
– model combination,
– or content structure
for this kind of SEO?

Right now every model eventually falls into the same repetitive “AI luxury voice” after enough pages.