After reading your question text, it Sounds like you're already asking the right questions, so you probably don't need my help to think about this. But I was looking for Qs to answer and hey why not.
On reading the title only, what I thought was: Yes, of course there is a business opportunity. And the reason why they are "so bad" is probably something to do with, the "learning function" is very inefficient.
By that I mean, what sort of feedback does one get on a JD? How can you measure a good JD versus a not good JD? How can you "A/B test"? As they say, if you don't measure it, you can't improve it. I think the lack of a tight feedback loop in JD contributes to making their improvement very slow over time.
Of course, I could be totally wrong about this. In my experience, tech job descriptions are not so bad, but other job descriptions ( anything non tech ) are awful. What are they even talking about? Of course, that might be since they just do not apply to me.
Whatever your idea would be -- I have an inkling ( not that I really have any idea, but hey ) that it could not simply just be another job posting board. It would have to be some other sort of business model. Maybe "A/B testing" for job posts? Anywhere, that probably already exists.
Finally, what just occurred to me is: HR / recruitment, it's very much "its own industry". They "have a way of doing things", and are sort of "shielded" from outside, by having a "monopoly" of sorts on the service they provide. Making it seem to me like they may be slow to adopt your "new fangled way" of doing their core business activity. Of course, I could have no idea what I'm talking about. Good luck! Sounds promising.