I started using pixate a couple of weeks ago and I hit some limits here and there,though these are two I hit all the time.
Though, these are just some quick suggestions, but I'd say they are enough to start with.
1: Spring animation does not allow to set a timing limiter. It all requires to be timed by friction to force ratio, which is pretty vague. Even though CSS offers no spring physics, it is easier to fake a bounce with chaining transforms in an animation sequence in CSS than to get this vague mechanic working right in pixate.
Suggestion: Adding a CSS engine for snippets - maybe js, which would make pixate come close to framer
2: Chaining animations in general is pretty hard to follow. You need to hack delays to chain animations. This is okay, but there needs some informative feedback system to show you where when animation fires in a heavy delay pack. At best you can also set the delays rapidly and in a comfortable way
Suggestion: A chained animation list tile that displays the sequence of every animation for an item.
3 . There is no way to copy an item's animations without cloning the item itself. This is a huge issue, that steals a lot of time. In framer a function does this all on its own, but in pixate you can not even copy animations. You have to look back and forth to copy the data, piece by piece.
This stole so much time... it is sad.
I am a huge fan of pixate as there is just framer which can deliver interaction animations on this detailed level. Everything else is way inferior like say marvel, principle and others and quartz is simple to loaded.
Maybe google gives you some room to work on pixate again some day.