Creating tailored products for people is the first step of R-tech. The second is recalling their preferences intelligently. The third step is anticipating what they'll want even before they articulate it. That's a measure of any great relationship. You can boast you really know someone when you can say, "I know she'll love this book!"
It's remarkable how well this simple system works. I eerily recommended great albums that I liked. There are many refinements to increase its power. I can "teach" the system by grading the results it gave me. Perhaps it recommended Pete Seeger because I named Bob Dylan as a favorite. But say I happen to already know Seeger's work and can't stand him, so I tell it to forget Seeger (and thus Seeger-likes). It's now smarter. I can further locate my space with more precision by rating as many albums as I wish, indicating my love or hate of them. (A strong negative rating is just as useful as a strong positive rating.) Because it is the web, I also have the option of listening to music selections to refresh my memory or evaluate recommended candidates.