When it comes to discovering/judging opening theory, I'd prefer using databases to engines. History tends to be a bit more accurate about how good a line is ... unless you're exploring a possible novelty.
To answer your main question => outside of a good theory book : annotated games, even non-annotated games (via databases) should be the raw materials you need to develop your repertoire.
The Slav is heavily studied, but for non-titled levels of play, just going over a ton of high-level games (both won + lost by Black) should give you a good feel for where the pieces go. I did that when learning the Scotch Gambit (heavily documenting whatever I learned from history's games)
I want to start playing the slav as black but I don't know much (if anything) about the opening. Would it be possible to build a reliable repertoire without having to buy a book on the slav defense?
In one of my other opening books I get some great insight into certain good moves, or the author gives practical alternatives. Once I ran one of these through houdini and it didn't like the suggestion... until about 7 moves had passed and then it finally agreed. I'm worried if I try to use my engine that it will lead me astray unless I give it tons of time to think... and even then I get no help on the practical alternatives.