I'm looking at this as a long-time FH user, an unwilling Illustrator user, and the former Sr. product manager for "Virtuoso" - which was the Windows NT/NeXTStep/Sun Solaris version of FreeHand.
I would start simple - get the Bézier curve editing tools right. All the little things - being able to use the keyboard in conjunction with the mouse to modify handles and points. Get that done, and I'll use it, even without the bells and whistles of Illustrator.
Pay close attention to the U/I and usability. KISS (Keep It Simple, Stupid!) is a great philosophy to live by. Less is more. Make it as simple as possible, but no more. Even FreeHand had some glaring problems in UserInterfaceLand. Fix them. (For instance the disconnect between where you'd access line controls and the ability to apply effect to lines.) Think like USERS. Not DEVELOPERS.
Prioritize everything. Speaking for myself, I could care less about all the 3D stuff that was in later versions of FreeHand. As long as I can distort an envelope, I'm good with it.
Look towards a modular architecture. It would be insanely great if Expressive were written from the ground up as a framework, where EVERY tool would be a "module" that you could add or delete. That way, a user who doesn't need a particular feature could remove it, making the app faster and more responsive. Thus, Expressive could be "tuned" for use by print guys, technical illustrators, and webdev guys - each who have different needs and perspectives. Other users who want more features could add them. And if you open your API, this could lead to a robust market in add-ons.
Lastly, get users involved, early and often. Set up a dev process where user feedback plays an important role. Don't fall into the trap of hubris, thinking that because you know how it's written, you know more about software than a user does.