Organizing Reference Library - Filesystem; DevonThink; Other Options?

I have a question to throw out. I’ve been trying to determine how to best organize my reference library. I currently have a very well classified filesystem. Most of the documents are in PDF, but I’ve seen a lot about DevonThink and other potential options. My primary goals for the system would be organization, ease of access, and being able to capture/create some metadata about the materials themselves (currently I just do that with tags, file attributes, and in-document annotations).

I’ve struggled with understanding the benefits of DevonThink, and the best answer I’ve seen thus far was from @OogieM who posted the following in the OmniGroup forums.

Some things that I still haven’t figured out with DevonThink are:

  1. Is it better to index or import documents?

  2. If you import documents, how do you open them from an app? I can’t open them (to my knowledge) in MS Word (for example) or PDF Expert from the open documents dialog box.

I have contemplated whether just using reference software on top of my file system would be able to provide the additional layer of organization that I am craving. I’ve used Zotero for a while, but not in any systematic fashion. I did find the following discussion helpful, but the academic focus does not quite apply to my use case:

One other consideration that I’ve had of late is just dumping all the materials in to Apple Books and creating appropriate collections. But some of this material gets dated over time and it seems easier to delete a file, but maybe I’m wrong about that.

What I really don’t want to do is (1) add more steps to my workflow; and (2) add more layers and busy work.

Appreciate any thoughts. Can also describe more of what I am doing if it helps.

Hi, first I know you can “open with” if you right click on the desktop, not sure on DEVONThink To Go

1 Like

I use Bookends as my reference manager. All the PDF files are stored in a folder on iCloud. That folder is indexed by DEVONthink.

I use indexing for files that I would normally edit in some other program. Mutable file’s, if you will. So the above mentioned PDFs I can view in Bookends, annotate in PDF Expert, etc. As if DEVONthink weren’t even involved. Now if I open DT, I can use the AI to find relationships between those PDFs and other documents I’ve indexed.
I also have DT index the note files in my zettelkasten so that those relationships (to other notes or PDFs) can be revealed if present.

When I have things to archive, I store them directly in a DT database. For example, I send a copy of license numbers and receipts for software that I buy to DT, and those are actually stored in the database. Receipts for tax purposes go there as well.

So things that I work with or on get indexed, things that are archival get stored in DT.

People do a lot more with DT, such as creating RTF note files related to other files (sometimes called sidecar files), etc. An example would be if you had a Keynote file, say, SummerPitch,key and notes about that presentation in SummerPitch.rtf.
I mainly just use it as outlined above.

If you like a more visual interface to your artifacts, have a look at Curio. It’s a very powerful tool, and allows linking to files or embedding files within. These can be arranged in an idea space (I.e. Canvas), along with lists, index cards, connectors, arrows, etc.

I’ve recently been auditioning The Brain, an interesting application that helps you think about and discover relationships between things. It allows attachments, timelines, etc.

Tinderbox is interesting, but it somehow doesn’t stick with me, even after I buy updates. @beck has done some excellent videos on it.


I have a system that I’m really liking so I’ll share a bit about what I’m doing even though it’s full of redundancies and inefficiencies.

  • I collect anything and everything that looks at all interesting with Paperpile. I love using this tool and I’ve never loved anything that began with “reference” or ended with “manager” before Paperpile.
  • I sync Paperpile with Devonthink (import bc I lost all my PDFs once and it will never happen again).
  • I search Devonthink for phrases just like I would search Google Scholar, but I’m searching my own curated list of interesting articles. This helps me see connections between all the things I’ve collected in a way that’s far more relevant and specific to my own research. I search for people, citations, keywords, and quotes this way.
  • I decide what to read using the papers I find in Devonthink (usually with the Table of Contents feature).
  • When I read the article, I add it to Bookends.
  • I use Bookends for reference management while writing (if I’m not writing in Google Docs, if I am then Paperpile). I also use Bookends integration with Tinderbox.

Happy to say more. Also :wave:@JohnAtl.


Oh, posted too soon. I also…

  • Read using Devonthink or Devonthink To Go depending on if I’m on my laptop or iPad so that my annotations and highlights sync.

If you’re planning on citing the lit. in the PDFs you are much better using a proper reference manager than trying to shoehorn something with Devonthink.

1 Like

The DevonThink forums (recently switched to Discourse) are the best source to search. I prefer to index. The only downside is when your documents reside on an external storage. Then, in order to access the files, you will need to have the external storage plugged in.

I sense that you would gain no additional benefit by an academic type reference management app (Bookends, EndNote, Papers, and the like). You may gain something by Zotero since it is a bit broader in scope. The workflow summarized by @beck using Paperpile may also be to your advantage.

I do not find any of the “book management” apps (Apple Books, Kindle, …) to be of any use to manage books as resources. I find these apps are designed for reading books in a mostly linear fashion, not about culling resources from books (or PDFs) in a non-linear fashion.



I typically import rather than index. One exception is the database that holds my Lambtracker SQLite queries which are indexed because the files themselves live in a Git repo area on my hard drive.

My indexed documents are just text files I just open them in DT no worries.


That’s the hassle.
DEVONThink excels when searching for information, but I need to reference/cite exactly that stuff. So I am using a mix of DEVONThink+calibre+Papers3. It’s a Rube-Goldgerbian contraption, but it works and I don’t see any alternative… DEVON people: please include references/citations with BibTeX support!

1 Like

A perusal of the DevonThink forums back through history will lead you to the conclusion that this ain’t ever gonna happen.



I think DEVONtechnologies’ design philosophy seems to be to be very limited in reproducing features that are available from other software. The editors in DEVONthink are very basic. (E.g., the markdown editor is a joke.) No mind mapping, not reference management, etc.

I’ve always found DEVONthink to be a good hub for my work, but rarely the place where I create or edit my documents. And, for me, that’s just fine. I can integrate DEVONthink with anything I want.

The days of “everything” multi-tool software development passed by a long time ago. Too expensive to sustain; too complicated and it dilutes the core business.


Nor will opening a PDF in your preferred application with a single click.

Never say never :slight_smile:

Or use ⇧⌘O on the selected document. (And, for extra credit, configure BetterTouchTool to make a double click action in DEVONthink perform that shortcut.)


I am swamped at the office today, but I wanted to say how amazed I constantly am by how helpful and engaging that people are in this community. There is a lot of great posts here, and I’ll respond as soon as I can. But I wanted to say thanks – thus far – to @Jonathan_Davis, @JohnAtl, @beck, @dfay, @DrJJWMac, @OogieM, @Lars, and @quorm. I appreciate your time and great feedback.


As a point of reference, I’ve got Curio (for layout and brainstorming), Papers3 (for the local citations and further searches), Mendeley (for group sharing of citations), and BibDesk (for the LaTeX bib databases). I’ve not managed to shoehorn DevonThink in this mix (yet), mostly because I find it is wearisome to navigate and/or because I believe (and grew up with the practice) that one should take notes linearly while reading a document and review them cohesively afterward rather than use a search engine to collect specific needle-points from a pool of as-yet-unread documents. I am also looking to move my reading/reviewing/annotating of literature to an iPad using an app akin to MarginNotes or LiquidText (when I would blog, I’d have more to say about my preliminary findings for this proposed workflow … perhaps this might show up instead here at some point).



@quorm: Is there a way to convince DTPO to always open PDFs in a specified app? I tried setting it at the OS level and DT still thinks I want Preview.

In my experience, if I select a document in Finder and open Finder’s Get Info, then choose the app I want as default and “Change All…”, then DEVONthink picks up the change and respects it. The first item in DEVONthink > Data > Open With… should be the app you defined in Finder’s Get Info.

I’ve never see that fail. Sorry.


I use DT as database for “finished” work and reference to be used in Agenda and locally saved Ulysses database. I work on a document in Word, finalise it, make a PDF and Hazel puts it into DT. DT syncs with DTGO and I have all the necessary resources at my disposal (e.g. to be used in LiquidText or callback urls that work both on Mac and iOS). Work in progress that will be split between Mac and iOS has a special database in DT and/or is located in the cloud secured by AxCrypt. I only wished Ulysses had a secure sync (not iCloud) so I could use it more.


@JohnAtl - Thank you for this and for drawing @beck into this post. I think the benefits of DT incorporated into a system where it’s as

May be very useful to me. I don’t really want to let go of my PDF Expert annotation process.

This is very helpful. Thank you.

@beck - you are one of my new favorite people. I’ve scoured your website and while my current system is different than yours in some respects, our foundations seem identical. The ability to search my own curated list the ways you describe is important to me. Thanks to you and @JohnAtl, I’ve now spent hours reading about this curious note-taking system known as zettlekasten. It’s fascinating. I’ve been a die-hard outliner since I learned about them in fourth grade. But, again, the process I use is similar to what you zettlekasten practitioners do. I capture notes in an temporary place and then capture them permanently into an appropriate home in my outline. Anyway, I was encouraged to hear that your very well-thought out system has redundancies and inefficiencies. I, too, can live with those, as long as the system itself has minimal friction points. Your posts and your website are extremely helpful. Thank you.

@DrJJWMac - Thank you for understanding the “problem domain” and your synthesis.

I think this was right on.

@OogieM thank you for sharing more about your usage. I am starting to feel like, if I use DT, a mix of imported (archives) and indexed materials (like the mutable content @JohnAtl mentioned) may be the right solution.

Thanks again, all of you referenced here or not, for your helpful contributions as I work through this.



Have you rebooted since the OS change?