What if you inverted the product idea and started with an online object repository that was presented as an API for search, retrieval and instantiation. You would include the basic design methods for parametric designs in your API, but you wouldn't be directly challenging AutoCAD's desktop business.
It seems like the missing piece these days is not the design tools, but the marketplace and the process of taking a design and validating and manufacturing it. If you could network a number of machine shops of varied capabilities and material specialties into a single point of contact for rapid assembly of completed designs with a marketplace for designers to sell their work, that could be a powerful combination.
There are already a few companies dabbling in this space http://www.shapeways.com/ being the one I'm most familiar with. The most interesting approach would be if you could find a market niche that had idle capacity and a need for creative custom solutions that was currently going unmet.
I'm not sure how that would work, but anyway I was talking about architecture and CAD for that only. How would you design something in your proposal? By writing code to glue together a bunch of objects? You need WYSIWYG editing, otherwise people could just write povray scripts and do everything locally. Maybe I'm just not understanding what you're proposing?
I'm thinking of something that acts as a catalog of pre-designed parts, that you can import into autocad, CATIA or even blender or maya. You could probably even put together a web-based interface to the geometry. It would act to some extent as a translator between different formats (.dwg,.stl,.vrml97, etc.) but the main purpose would be to act as a marketplace for designers and production bureaus. The ideal use case would be a construction project manager who's been tasked with remodeling a space on a tight time budget and shops for custom fitted design elements purchasing several designs from different sources and ordering production from one or more production bureaus within a 1-day delivery radius.
The real chicken and egg problem with this idea is that you won't have enough designers and qualified production bureaus to form enough of a critical mass of purchasers until you have a critical mass of purchasers to make the service attractive to designers and production shops. It would probably work best if you start off with a small regional network of production houses that employed their own designers.
Ah I see, yes this is also part of the 'BIM grand vision'. It's in the final stages of implementing an industry-wide system though. This would be hard to do without a solid toolchain in the beginning of the process, with a client to visualize how those parts look/work together. I don't see this as a replacement for 'traditional' tools, more like an add-on.
Big manufacturers already provide construction blocks, in standardized formats, for importing into design tools. From ICF forms to wallpapers. The 'marketplace' part is missing though, that's true. I don't even think it's a matter of technology, it's just that the industry isn't ready for it IMO. Construction is a very conservative industry.