I don't know about treeline. At a previous company, we committed early on to using Parse and it was a disaster for us. Most of our problems as a company were not technical of course, but pretty much all our serious technical problems can be traced to us locking ourselves into using Parse, when it would have been 1000x easier to just put a server of our own making on Heroku and run a postgres database.

Our userbase was TINY (so you think scaling should not even remotely be an issue), but we had constant downtime with Parse; not to mention we observed between 1% and 3% of all requests to Parse failing for obscure reasons (and it was never clear if it was just our node, or if it was global). I seem to recall the Parse team fixing these, and then repeatedly rolling back their fixes since they would inevitably cause something even more catastrophic to happen. I am sure all of them were working really hard, and I don't want to say anything to make them feel bad, but honestly, the class of problems that plagued Parse (at least at the time I was familiar with it) did not inspire confidence. It is possible that it is much better now; I have not used it in months.

Parse is a lot of fun for quick prototypes and demos, and I recommend it for that if you find yourself wanting to iterate quickly and repeatedly on such things, particularly if most of your "innovation" is happening on a client (like an iOS or Android app, etc.).