Considering the map design of Larian even a big city will only be a couple of houses next to each other.
I'm not quite sure why you say that? You may well be right, but I'm not sure exactly where the tension lies.
To me, and admittedly the experience I have in software development has absolutely nothing in common with designing games, there's no obvious tension.
Take, for example, the druid grove. It has a bunch of NPCs including 3? merchants (Aaron, Dammon, Ethel, not including Matis and Mol), a big continuous multilevel area that's part of the main area map and four separated smaller areas (Zevlor's base, Mol's cave, the Enclave Library where Kagha is, and the underground passage). It seems to me that if you changed it to have different graphics to make the big multilevel area less cave like and more urban-looking, and instead of having so many NPCs out in the open put more of them in apparently enclosed buildings like the one where you find Pandirna (the injured tiefling) or the ones in Moonhaven, then changed the separated off areas to be something like an inn, a government building, and a couple of other bigger buildings (potentially with some combat-related questing going on) then you'd have the makings of a pretty decent small town.
And I don't see why that wouldn't be scalable to a reasonably convincing city, if you weren't trying to include a goblin camp, abandoned village, ruined temple and various other isolated buildings and areas on the same map.