The way most suburban residential construction is done these days (at least in the U.S.) is to start with a large area, split it up into relatively homogeneous lots, and flatten them out as much as possible so that they fit the requirements of one of some set of floor plans that were designed without regard for any specific lot.
The alternative to this is to do custom designs that take the considerations of particular lots into consideration, like the position of the sun during the day, the direction of the best views, layout of the yard, etc.. This is expensive and hard to do on a large scale, because you can't amortize the cost of the architect over multiple buildings.
Another alternative is to have a software package that takes a given lot and generates a floorplan optimized for that lot (and the needs of the buyer, if known). This is a hard problem, but I think it gets easier, or at least more interesting, when combined with machines that can build houses. With 3D printing, the design can be as complicated as you want, with little impact on cost. You can have funny-shaped walls that aren't at right angles with each other, vaulted ceilings, gothic arches, flying buttresses, or any kind of ornamentation or functional feature without regard for having to consider the difficulty of expressing the design to human workers. If done well, it could be a major step forward for well-designed residential construction -- you get a custom building on every lot that's optimized to match the site rather than forcing the site to match the builder's requirements, and the marginal cost of the design work is low (probably more than zero, since you'd need a human to review the design and possibly negotiate with customers).