Admittedly, there's a lot of people who don't really do the improv element as much as I do. It started out as a sort of "lazy" moment for me, but later on I discovered that it's a whole discipline.
The roleplaying community seems to be of two minds on whether it's good to obsessively prep or improvise, at least from what I've seen in my personal experiences. A lot of people are big on having really heavily scripted content, with everything in its place and preordained, which is helpful for some games (coughD&Dcough) more than others.
On the other hand, there are games that lend themselves to being run with much less prep, and even ones that focus on the improv method. The real turning point for me where I started to appreciate improv was Savage Worlds, which I only ever played once, but whose adventure frame format was highly influential for me.
The way it worked was entirely focused on drawing connections between main points, so the plot would progress in a predictable fashion and the GM could still do pacing and build up tension as one would for any story, but the players weren't funneled into any particular path of action, giving them the sense that they have a lot of agency in the universe.
Often my prep for games is so sparse that were I to write it down you'd just see a couple numbers and maybe three or four sentences. This can be a little ironic, since I've occasionally gone to a fair amount of effort to draw (mediocre) maps for my games, and I certainly value the storytelling above the visuals.
More power to you. The improv element allows everyone to buy in. I like the way you said it: [Savage Worlds] giving them the sense that they have agency in the universe.
There's probably a happy middle-ground between planning and improv, but the longer you play tabletop games the less planning you probably do. In improv no one walks on stage with zero prep, years of experience has gone into tricks and methods for heightening scenes. I'm sure the same applies for for tabletop role-playing games.
I checked out "Savage Worlds" from Pinnacle Entertainment and downloaded their 'Test Drive' rules. I'll play it with a few friends; if the first time is a wash, I'll use a 'Bennie' and try again!
To be fair, when I talk about prep I usually think of the stuff you do for an individual session. When you spend hours and hours reading through details from a setting, you've actually got quite a few tools at your disposal.
Typically, all I do is build a knowledge of the setting, then think of something that someone in that setting (that the players will interact with) want.
For instance, one session might be: A salvager wants to recover lost technology from a ruin.
The next session: The salvager and the commander of the enemy forces are working together on a plan to facilitate her defection to the salvager's company.
The secret is knowing how much stuff you have to do around that to get it to work. Having decent setting familiarity, encouraging questions, and giving little tasty morsels of mechanically or narratively interesting information is usually enough to go on, if you're comfortable with the underlying framework of the system.
This is why, for instance, I don't run D&D very often: the mechanics are very well balanced, but they're incredibly complex to reach that point, and you can very easily either mulch players or fail to challenge them. The solution would be to just nudge things on the fly, but then why do any prep in the first place? And if you're nudging things, it's also pretty easy to step on the players.
Games like Savage Worlds try to abstract out a lot of stuff, and that makes it easier to do on-the-fly changes and tweaks without feeling like you're just choosing the outcome beforehand. However, even a very mechanically complicated system, like Shadowrun, can have easy "baselines" to follow.
From a mechanics design perspective, you can find systems that are very picky about how you interact with them. D&D, for instance, has high player-enemy asymmetry in mechanics, and multi-stage attack/damage with arbitrary randomness, while Shadowrun is more generally about mechanical symmetry and performance-based results. The latter approach means that you can base things directly off of the player's characters: a weak enemy is just them but with worse dice pools. A strong enemy is just them but with better dice pools.
Of course, a good solution to this is just to avoid combat, since most games have fairly elegant non-combat mechanics, but counting on players to do something that's convenient for the GM is a recipe for disaster.
I hope you enjoy Savage Worlds. I personally liked it quite a bit, but my group wasn't as enamored with it. We were going through a phase where we did new games every couple weeks, and they just didn't click to it.