"The killer use of this, IMHO, would be the ability to integrate your technology into an already existing block of text hosted elsewhere just by dropping a line of js into the page"I double vote for this approach, for selfish reasons (i.e. I might be a customer one day). The reasons are clear: Getting to end users directly is expensive. It would be 'easier' to sell Revizr to vertical applications as a plugin and let them spend all that money.
However, that being said and being less selfish, I wonder if the world is ready yet to build web apps out of other web apps. I don't think so yet.
Scribd is a good comparison. They went ad-supported because that is probably the only model right now the market will understand, even if it is not viable. If the tables were turned and I was you, I would really want to charge directly for my costs like a proper subscription SaaS service. Here's a sobering take on Scribd's model:
http://www.mathewingram.com/work/2008/02/19/scribd-cool-feat...
Other comparisons are web-to-print (e.g. postalmethods.com) and web-to-fax (myfax.com) services, but people are more willing to pay for those because there are obvious hard costs for them.
I believe that your future is all about being a 'feature' rather than a 'product'. I suppose I'm saying you ought to explore this model alongside an alternate revenue stream that will keep you afloat.