This cannot be emphasized enough -- start simple and consider strongly before adding complexity. In almost every case, you can get 80% of the benefit of automating a document with 20% of the work, so get that done fast and then invite your colleagues to use it.
In most cases, your 80% automation will be a Merge turnsheet without any conditional automation or formulas, where end-users just type in the relevant text in each column, because in most cases the lawyers who are using it will know what needs to go in the document. They'll do what they're already doing, but now:
they'll do it a lot faster and avoid wasted time on repetition, redlining, etc.;
they'll be working from the firm's best template, rather than a random precedent; and
it will be easy for others to review because the points that vary are arranged neatly in the turnsheet instead of buried in a redline.
Maybe a trainee or a junior associate won't actually know what should go in some columns of your 80% automation, but they can ask their supervisor in a comment and learn. Maybe users of your 80% automation still need to make some tweaks to their documents manually afterward, but if it takes an hour-long drafting task and cuts it down to 10 minutes, they have 50 minutes of free time to thank you. And if you can build something that saves 50 minutes of time in the course of an afternoon, but it would take you weeks to build something that saves all 60, you are doing your team a disservice by waiting to share what you've built.
When you do want to add more automation to get closer to 100%, whether on the basis of user feedback or your own pride of craftspersonship, do one small thing at a time.