I spent a few years learning parametric modelling in Houdini, then in 3dsmax/Railclone and dabbled with Rhino/Grasshopper. While I think procedural modelling is great its highly technical and a very niche market. Most customers want plug and play, most companies don't have the resources to R&D technical issues for days/weeks because you want it to be 'perfect', they are happy to have it look good enough and get it out the door. This goes the same for most artists/customers who work within tight deadlines, they need things to work fast, look great and move on to the next part of the project.
This is why scatter software is successful like Forestpro, its easy to learn and powerful with few parameters. Its also easy to teach someone else in a timely manner. Procedural modelling tools like Grasshopper/Railclone/Houdini take quite a while to grasp, you need to have some interest in mathematics etc. Don't get me wrong lots of people enjoy a challenge but my advice for selling this kind of thing is to sell the components ready to go with instructional videos to help them replicate it if necessary or just use an array tool to fall back on. Also have some prefabs ready for them for different situations, like cloning along a spline, a closed shape, a few different variations. If its easy for customers to use then sure I think you could be successful. Keep in mind the people you are targeting might be a very small percentage of the market, so sales could be on the lower end - or maybe you are right its the future and you nail it on the head. I thought this in 2012, still waiting for that day lol.