DEVONthink with Zotero Workflow

I wish I would have used Zotero and DT for my dissertation, but it is already changing how I read and annotate research.

Check this thread out for a good starting place.

1 Like

I am curious why not DT and Bookends?

Thanks @rpallred

@smilingtiger, a number of researchers I’ve run into recommended Zotero, so that got me started on the journey. It seems like there are Zotero integrations for just about everything (Word, Scrivener, etc.). The other interesting option I’ve run into is Mendelay. I’m open to suggestions.


Thanks, actually after doing some research i have come down to Zotero and Bookends, I still dont know which one i am going to choose …

For me I don’t even remember if Bookends was an option when I made the decision. My graduate program gave us access to EndNote for “free”–so once I graduated, I needed something new, cheap (Student loans you know…), and cross-platform. At the time, I could run Zotero completely in Firefox on my work computer, and natively on my Mac (I don’t think that Firefox-only version is still available).

I also found ZotPress and phpZoteroWebDAV which allow me to incorporate all my references into my website.

It also plays very nicely with Word and LibreOffice–and at the time nothing but EndNote worked with Pages–haven’t looked again.

Zotero is terrific because it is open-source and has a very active developer community.

But there are limits to how polished an app can be and how quickly new features can be added with an open-source app. Most notably, Zotero has no Applescript support and thus a number of integrations are done instead via Bibtext and Terminal-type tools.

Bookends hits a really nice sweet spot as it is developed by a small (perhaps 1-person) shop clearly dedicated to this mission. While it is not open source, it has extremely nice capabilities to customize its functions. It is impossible to open the extraordinarily detailed Bookends documentation and not learn some new feature which you overlooked for years. On top of that, Bookends has Applescript support so that makes it a lot easier to integrate with other apps.

If you get into detailed analysis of academic literature someday then Tinderbox is the utlimate tool in that regard; for all the above reasons, Tinderbox integrates very well with Bookends but not quite as well with Zotero.

Mendeley, Endnote, and others are largely commercialized and not nearly responsive to customization/integration with other apps as Zotero or Bookends.

Given the above, personally I have long used Zotero and in fact have developed or commissioned a number of integrations between Zotero and other apps (see Zotero - DT3 Import Working: A Customizable digital word processor for HTML, RTF, and PDF ). But particularly given the increasing role I see for Tinderbox in my curation of academic literature, I am largely switching to Bookends.


I experimented briefly with referencing in Nisus Writer Pro – using BookEnds – and found it rather crude. By contrast, referencing in MS Word – using Zotero – has worked quite well for me.

I’ve sort of integrated DT3 with Zotero (not really), but would be grateful for insights as to the possibilities of DT3/BookEnds integration.

It’s built-in


I also use both Devonthink and Zotero. I use zotero to hold my pdf files and to categorize my sources in a single source folder without worrying about structure, which I can move around within zotero. I use Devonthink as my notes repository. I actually don’t index my pdfs in DTP3 because they are a bit too large and are held in an iCloud folder to save space on my mac. (Although I imagine indexing my pdf files in Devonthink would improve their usefulness!)

I’ve found Kourosh Dini’s taking smart notes with Devonthink to be helpful! It is definitely worth a read. It walks you through the steps of creating a note, creating a link to zotero, and linking notes.

1 Like

Thanks: I don’t think I’ve looked carefully at that pull-down menu for several updates.

That said, I’m not really sure about the utility of importing Bookends metadata. Would rather be able to export my DT metadata into a cite-able format.

Thanks: I’ll try to find time to have a look at Kourosh Dini.

I have a similar set-up, using Zotero to store journal articles, ebook .pdfs and other items where metadata can be downloaded directly into the app. Most of my note taking in DT3 includes internal links to items stored (or indexed) in DT3. It can, however, get very fiddly and time-consuming.

I use Zotero with DT. This is because I need to work across platforms with people who may or may not invest in software. I can fund the cloud space and work with groups. Zotero provides an easy way to share core resources with co-authors, and pull a formatted reference. However, a reference is a likely to be in Scrivener if I’m note taking on something specific I’m writing. DT is my go-to for storage and retrieval over time. I’ve always wished DT would save me a ton of effort by simply adding the reference management functionality. I almost never do an end of project substitution of references for tags. Much easier to use the A-D formats and just do them as I’m writing. Zotero just gives me an easy way to do the first-time citation and build a project bibliography. IMO the days of “ob cit” are over. In fact, often the URL is often important for the convenience of readers, also for in-text citation. But I mostly do consulting writing and my readers want convenience rather than citation elegance. I would say that in some cases with document acquisition it makes sense to collect documents into Zotero first from those sources that will support the transfer of citation meta data automatically. However, so much of what I work with comes from sources that do not facilitate this, I often end up making a manual entry anyway. It’s amazing how many document creators don’t include clean document info, even, for a PDF.


This is sadly true, even for so-called professionals.

1 Like

I tried Zotero and there was much I liked about it, but, ultimately, the non-native UI was just too much friction. (The same holds true for Obsidian.) I use Bookends to organize my PDFs of articles and books, and to generate bibliographies for work I do in Scrivener. I take notes on those PDFs in DevonThinks because it is, to be honest, freaking magical.

The way to do it is not to import your PDFs: leave them wherever Zotero or Bookends places them – Bookends just keeps them in a single directory in the iCloud Drive directory (because I sync with an iOS device) and index them in Devonthink. Then annotate them in PDF and DT creates a new folder called Annotations which, if you have added the links that DT lets you create, will take you straight to the spot in the PDF that is referenced in the annotation.

This is as close as it gets to blissing out while taking notes.

1 Like

I am familiar with Zotero but have been using ReadCube Papers (still in Beta), which is destined to be the very best reference manager going around. And I’ve used them all. I would love it if Devonthink was able to create .ris for media articles and so on (for which I’ve curated pdfs) and export that information to ReadCube Papers. I have tried to do this with Bookends and then on forward to Papers, but it is a lengthy and convoluted process. If Devonthink had the capacity to create .ris (or similar) outputs for all file types (obviously with some manual adjustments as needed) it would be hands down the software of choice for (Mac user) academics globally. Because that would allow greater versatility between curation of content outside a reference manager (i.e. web pages, tweets, media articles, and so on), followed by the export of citation details to that reference manager in order to cite the materials in manuscripts written by academics. Reference managers do a good job of downloading academic papers with DOI numbers, but they’re very poor in terms of curating anything else. That’s where Devonthink is so brilliant at capturing. If DT could convey to my reference manager what the citation details are (using .ris or similar) it would be a workflow dream.


Interesting - what features are upcoming from the beta version which have you so enthusiastic?

It’s actually the first request of this kind that I’m aware of. Is anyone else interested in this? What exactly should be exported and how/when/why would this be useful for your workflow? Thanks!


It would be useful to me because (1) RIS is a common denominator of much reference management software; and (2) Tinderbox can input RIS data and incorporate that into its custom metadata, which it terms “Attributes”

That said - @DarylAdair have you experimented with the existing “Create Metadata Overview” feature? Much of what you want to achieve with RIS can already be done with custom metadata exported into a TSV or CSV file via Create Metadata Overview. And much software that exchanges data via RIS can also exchange data via TSV/CSV. So I am not certain that RIS support is needed given the flexibility we already have from that.

1 Like