To elucidate an example, imagine that you have to build a recommendation system for keyboard design, where the system should not only use previous designs in the dataset but also suggest modifications from previous projects. For example, a mechanical keyboard is in the dataset, but the output has a mechanical keyboard with RGB LEDs. Assuming that we have a set of rules that suggest plausible modifications to designs. Since each design depends on the team of designers, each project has a different output depending on the designer's choices, and therefore, I does not seem to be compatible with a collaborative filtering approach.
Search for Knowledge-based or Constraint-based recommender systems. Quoting the "Recommender Systems Handbook":
Traditional recommendation approaches (content-based filtering  and collaborative filtering) are well-suited for the recommendation of quality&taste products such as books, movies, or news. However, especially in the context of products such as cars, computers, apartments, or financial services those approaches are not the best choice (see also Chapter 11). For example, apartments are not bought very frequently which makes it rather infeasible to collect numerous ratings for one specific item (exactly such ratings are required by collaborative recommendation algorithms). Furthermore, users of recommender applications would not be satisfied with recommendations based on years-old item preferences