If your work should be compared to a software tool, then you should compare it to that software tool. Actually comparing, however, can be tricky for a couple of reasons:
- Closed commercial software often hides its workings, such that it is difficult to tell how it actually operates.
- Live software projects are a moving target, and features today may be gone tomorrow.
Thus, for example, if you were comparing a contextual text search to the methods used by Google, it would be very difficult to do that just in a "related works" section because their methods are hidden and constantly changing. Instead, you'd need to take experimental data on its behavior in some period (or cite somebody else who had).
For this reason, whenever possible it is generally preferable to reference a paper about the software rather than the software per se. Most significant academic software will have an associated website with a section on "how to cite this" or a list of associated papers (e.g., this "further reading" section for TinyDB), which will give you appropriate sources to cite in your discussion.
If there is no such citation available, however, or if the existing citations are not appropriate, then you can cite the tool directly: there are established conventions for citing software tools. I also sometimes cite both a key paper and the live tool, when the tool has expanded well beyond the associated literature.