Have you thought about easing distributed collaboration for structured data? The use case I'm thinking of is multiple groups submitting data to one central informatics/analysis group (i.e. consortia, meta-analyses). Scientists still throw around data in excel sheets with ever-increasing versioning through file names (e.g., gene.data.v29.201230429.janes.final.Final.reallyFinal.xls). It's a huge time-suck that doesn't help anybody. Support for easy viewing of XML, JSON, or even some sort of hosted RDBMS would be incredible.

I think it's great to target the wider, non-developer audience. They could benefit immensely from version control, distributed systems, and smart editing tools. It could be a big benefit to the whole scientific community.