I have had DevonThink (DT) for five years or more. I struggled to find a serious use for it until recently. I am now using it to cull documents collected from over two or more decades of teaching courses. My goal is to create a reference library of lecture slides, worked examples, demos, study guides, study problems, and additional resources for each topic in a given course. To that end, I have imported the source documents into a DT database and am working methodically to mark (and remove) duplicates as well as to categorize and sort the remaining documents (by topic followed by type).
Given this objective, I have not found any other app that allows me to work through my resources with the effectiveness that is provided by DT. This is not to say that DT could not do better (and I direct my comments to this end to the DT Forums). But, after 5+ years, I am finally appreciating the need for DT in my toolbox.
You did not ask directly, but implicit in your question might also be the inverse … “Why NOT DevonThink?”. To problem raised by @OogieM on data loss: I’ve not experienced it. You might decide whether her warning means that you should avoid DT entirely or whether it means simply that you should develop a stronger contingency plan for the potential that data loss could happen to you.
My one reply to the “NOT” based on the 5+ years of experience is: Do not get DT if you are thinking that you need it to be a more sophisticated way to do Finder-level actions. While DT is more sophisticated than the Finder in its ability to search, I think the overhead is too high to use it on a routine basis as simply a better Finder tool.
My one reply to the “NOT” based on recent advances in apps: Do not get DT if you are thinking that you want it to help you connect ideas across existing or as-yet-to-be-assembled collections of documents. Up to the point of Roam and Obsidian and comparable tools, DT may have been the best macOS app for such work. In the meantime, I am finding that, compared for example to Obsidian, the overhead is too high to use DT as a tool to generate and investigate connections between document content. However, see also the note (caveat) in the next paragraph.
Finally, what about the possible limitations with Obsidian (or Roam or others) compared to DT in the ability to search inside documents to expose the content that can then be connected? I take a neutral stand on this question. First, I am entirely naive about whether Obsidian or Roam or other (markdown-centric) apps can or cannot search content inside documents (e.g. PDFs). I cannot imagine that they absolutely cannot (or that eventually some plug-in will allow the search to happen). My comparable work here is that I read PDF documents in the science/engineering discipline for research developments. In this realm, I have a dedicated database app (Bookends), and I do the work manually (by annotating the PDFs as I read them). To continue the note in the previous paragraph, I am also exploring how to bring the annotations from the PDFs into Obsidian rather than DT. This said, I am aware that other folks post accolades about the ability of DT to search inside documents, reveal the content, and expose otherwise unseen correlations among the revealed content. So, as the saying goes … YMMV here.
Hope this provides some useful insights.
–
JJW