DEVONthink + Obsidian

My workflow is very simple. I basically use a variant of the PARA method. I am in the academics.

3 main folders in my Devonthink:

1. Resources:

  • Notes I accumulated over the years, on my academic field
  • Pdf articles and books I collected over the years: they are managed by Bookends and indexed in DT. Some of them are in Zottero. They are also indexed in here.
  • Scapple files
  • Any other material in my field.

2. Projects:
This the folder I spent most of my time. There is no real file in here. Everything here is virtual. Files and notes are collected using smart folders to appear within projects.
-I am a linguist. So, assume I am working in a specific topic such as causatives, I would first have a folder by the name of the project: pCausative.

Within that folder, I will setup a smart search based on a number of criteria.

Two steps here:

a: Collection stage: collect everything that could be relevant to the current project:
Smart searches will aggregate all the files. The files from the Resource folder will appear here.

  • Causative[tag]: this folder will collect all the files (both pdf and md, and rtf) with the tag [causative]
  • causative [content]: will do search a content

The collection stage also involves searching files using Foxtrot. I get better results with the proximity search in Foxtrot. Once I do the search of the relevant terms there, I assign tags on the files that I want them to show up in the project folders.

  • You can also use Finder’s search to get relevant files. What you need to do is assign the relevant tags. All those files will show up in your project folder within DT. That is the beatify of DT. It works with any file type. Tools such as Obsidian are restricted to plain text files. You have no way of aggregating, filtering and working with a lot of file types.

The searches will definitely over-generate: in the sense that many irrelevant files will be collected within in the smart folders. That is why I need the filtration stage.

b: Filtration stage: I have a special tags for the specific project. This is not a regular tag. These special tags start with p_. They are temporary tags that I will remove when I finish the project. During this filtration stage, I assign those special tags on the files that I want to use in the project. I will go through all the files that are collected by the smart folders, and assign the p_causative tag. Files that are not going to be used will not get the tag, and the smart folders get deleted at the end of the filtration process.

Once I have done that, I am ready to work. I am going to read the articles, and notes I have about the topic, compare the ideas there. If there are other relevant articles that my searches missed that I need to read (based on the citations on the papers), I will go and assign the _ tag to them.

As I read the materials, I will develop a hunch or proposal for the problem I am working on. If the ideas are complicated and hard to workout, I will move to Scapple to get the visual aid. Otherwise, I just start to draft the paper there, within DT.

If the writing is going to be long and complicated, I might move to Scrivener as well. But, in most cases, I do my first draft within DT; and then move the draft to Latex for final cleanup .

3. Archives

  • completed projects are moved here

The fourth folder in the Para framework is known as AREAs. I honestly don’t much care much about other stuff in my life. I just use a simple Evernote library for reminders, for remembering stuff, short notes of shopping lists etc. I use Evernote because I can access it on my phone.


This is a great use of DT, thank you for sharing it!

Terrific. This deserves a thread, if not a “pinned” post or even a DEVONthink blog, of it’s own, unrelated to Obsidian. @BLUEFROG

1 Like

Hehe I was thinking the same thing, @Dellu needs a DT blog article. This use case works beautifully for academic papers, but I suspect would also work for in-depth non-academic writing (e,g. Investigative journalism).

It’s also actually the first time someone has inspired me to think properly about smart groups. I like the idea of a “separate section” of my database which is “calling up” the files in my groups, but arranging them in a different way without affecting my “primary” structure (as mentioned elsewhere I have a “rigid” folder structure that I like and don’t want to mess with).


This is really interesting. I’m a researching academic in politics and I have been stressing for years how to organise notes. I tried working with Finder, Zotero and Scrivener during my PhD. It was difficult, but I got through.
Now I am trying to get on to Obsidian integrated with DT. But for the last few days, after amassing a hundred or more files, I’ve realised there is so much friction working with Obsidian (do I take a note in that program, or should I use the annotations in DT, etc).
But today, after spending an hour bashing my head against the wall, I also realised that the linking is not so helpful as I thought it was. Your post here is very timely.
I think I spent about an accumulative 4 weeks of my phd trying to work out if I should link my notes through the reading note document, topic document, or another mixture of both. I spent countless hours trying to perfect a template note for organising all my reading notes etc, thinking the links will help me connect the ideas. I lost so much time reading stuff from trying to organise how arrange future notes that I never ended up taking.

As such, I am seriously thinking of dropping Obsidian and all hyper links in note taking.
I like the tag method and some hierarchical folders in DT. It’s simple and it forces me to just read. Like my supervisors and all the other elderly faculty, who had to learn a whole bunch of stuff without hyperlinks and second-brain tools!

I also have a tag system for projects. I used to have “literature” folders in each research project, then replicate the documents into each project if I needed to, but I realised just tags with the project number applied to the sources in the main Literature folder is sufficient.
I’m curious as to why you delete the tags when you’ve finished the project?

I’m also curious, how do you take your reading notes - hand or computer? If both, how do you go from one to the other. How do you organise your digital resources with books and printed materials?
And how do you organise your notes? It seems like you read the sources first, take notes, and compare them with other sources. Do you then start writing a topic note with notes from your relevant sources?
Maybe it’s different with linguistics but I was always encouraged to look for information about the topic in the source, otherwise what’s the point of reading it if it doesn’t relate to the topic of your project.


Dear @rmschne and @MsLogica, thank you for your kind words. Appriciated!

1 Like

The suggest is noted.


Think of the most influential and successful scientist in your field. Do you think that they spend a lot of time creating hyperlinks across their notes, or trying out the latest gadget or app?
I don’t think so. A working solid system is what we need. It is not running with the latest hype.

In the Obsidian forum, I equated the Zettel method with the Tiktok. These are shallow tools: made to satisfy the urge to navigate.

I have tried to check the habits of some of the most successful scientist. Most of them never mention the methods they use: rarely care about them.

The two exceptions I am aware of, mentioned in interviews is that of Chomsky’s who said that he used the same set of computer programs throughout his life. He said his grandchild configured his computer in 1970’s: and he never cared to change any of it. Every time he changes departments, the department tech has to call his grandchild to move the old scripts to the new computers.
That is it! He is one of the most successfull scholars alive. He wrote all that material with some simple set of computer scripts that we setup on 1970’s. Never cares about the latest Ms word, latest Macbook pro. He spend his time on the productive thinking.

The other popular case is the renown physicist Richard Feynman. He quintessentially used just paper and pen to think about his problems. He said he keeps a list of his favorite physics problems in a paper. As new papers and theories emerge, he tests those new theories against those problems if they can work or solve them.

That is to say: most successful thinkers use very simple methods. Complex methods are obstructions for successful thinking.

Having said that, to answer your question:

I’m curious as to why you delete the tags when you’ve finished the project?

That is to declutter the list of tags in my system. Since the project is finished, those project tags have no purpose. I don’t want to see them listed in my list of tags. I am moving on to other projects. I delete them when I move the project folder to the Archive.

I’m also curious, how do you take your reading notes - hand or computer?

I use both. If I am reading and thinking on paper, and that I have figured out everything there, I just copy the major upshots from the paper and develop the draft on the computer.

But, most often, I do my reading on the iPad, sync back to mac to export the annotations. Writing comments in the iPad is not effective. For that I use paper and pen to write comments/notes while reading on the iPad.

P 2: H3: (page 2, 3rd highlight): then, write my elaborated note on the paper. I will copy the note when I move to the computer. If the comment is very short, I will write it on the pdf itself.

If both, how do you go from one to the other. How do you organise your digital resources with books and printed materials?

I don’t have many hard copies. I mostly use soft copies. I use Bookends to organize my reading material.
I have 2 libraries in Bookends: @ library which contains articles and ʙ library which contains books and dissertations.

The important part of that to make your pdf files contain the Citation Keys. That is, I configure my library in such as way that the pdf fiels are named in this patter:
Key Title. The Key itself is made of Author+year.

The function of this setup is that you don’t need a citation tool to insert your references. You have your citation key on the pdf file name itself. So, when I am writing in DT, or Scrivener or whatever software, I don’t need to worry about citation tools. I simply insert the key from the file name of the pdf laying in front of me.

And how do you organise your notes? It seems like you read the sources first, take notes, and compare them with other sources. Do you then start writing a topic note with notes from your relevant sources?

I don’t organize my notes. I simply dump them in one folder, and tag them. I write a simple summary at the end of the notes just to make the core ideas available for comparison with the ideas present in other papers. Yes, I always start by reading the works from other authors–articles and books relevant to the specific project. Once I figured what other people have said about that topic, I will think about the issues they never tackled, and what I can do about those issues.

  • There are times when I don’t read a paper from the beginning to end. I simply pick an interesting part or section, take notes from that section and move on. Those snippets of notes are also dumped in the same folder to the other notes. For those notes, I write the source.

  • Another habit I have developed over the years is to use different text colors to differentiate paraphrases of other works from my own ideas. I use bluer when I paraphrase other authors, and blue for my own elaboration, or new extended ideas. This is to avoid plagiarism: a serious problem in the academics. This is one of the reasons why I like the rtf format, and dislike markdown. You cannot color texts in markdown, unless you go through a lo tot of CSS stuff which will fail when you open it in another app.

Maybe it’s different with linguistics but I was always encouraged to look for information about the topic in the source, otherwise what’s the point of reading it if it doesn’t relate to the topic of your project.

  • That is true. Reading the works of others, you are going to learn not just the information there. You will also learn the methods of argumentation, organization, terminologies and other methodological tools that will be very instrumental for your writing. I have a dedicated tag for papers that are inspiring in terms of organization, logical argumentation, language clarity etc. I sometimes open one of them while I am drafting just to get an inspiration from it. We know Style is very important for successful writing. For that, having a well written paper on the side just to get an inspiration can be helpful.

I feel you. I’ve spent the last few months in the same situation, trying plugins and dozens of other programs, including shiny and expensive PKMs, starting with Obsidian, and then Roam, Mem X, Craft and several others. All to no nothing. I am also a researcher (Humanities) and many of these are excessively business oriented and very little futureproof, not considering the amount of files, abstracts, books I work with daily. In the end I came back to Obsidian for a single reason: working exclusively with .md files on my computer is the one that better integrates with DEVONthink. Yes, because the second thing I’ve noticed is that DT is already sufficient, they try to complicate a lot but DT is a very complete and self-sufficient program. We have to exploit it.

I haven’t gotten to a setup that completely convinces me yet, but I’ve streamlined to this:

  • DEVONthink. Where I store all sorts of files, it is the core of all my work. Inside it is a database with my indexed library, there are my markdown notes, personal documents, paperless office etc. The structure is strictly organized in folders following (loosely) the indication of Tiago Forte and his PARA method: there is a folder for each project I work on (containing drafts, contracts, notes etc) and an archive where I accumulate finished projects.
  • Bookends. My reference manager. I mainly use it for searching and entering citations when I write essays or books (I use Mellel, which integrates nicely): I do NOT use Bookends to read pdfs, for that I still use DEVONthink (but with which the library is indexed). Bookends is also organized in folders with the same name as the DT projects above.
  • Obsidian. Possibly the most superficial and least integrated part of my system, I use it essentially as a frontend for notes that are in DEVONthink. I don’t use any tags or complicated structures. I only have inside it 3 folders: 1) Notes (where I write anything, short or long, feelings about my day, research ideas or interesting things I think after consuming any kind of media. It’s not important that I tag them, I find it a waste of time: in case I need them for an article, I simply rely on DEVONthink’s AI-powered search or graph. 2) Annotations: single notes completely related to a book/paper I’m reading. Simply renaming with Author Data, and inserting a deeplink to the DEVONthink file in the first line. One book, one note. If I need to insert some citations from books in my library I simply write Author Data and copy the DEVONthink deeplink (not Bookends, the core remains DT). To do this quickly I use Raycast’s DT extension. I do not use citations plugins, citekeys o bibtexts. 3) Readwise, a list of my highlights. Why do I export it to Obsidian? Readwise does it automatically and then I can use DT’s search engine
  • Readwise Reader. I use it as a read-it-later app of articles found on the internet, the things I highlight within it sync with Readwise which are automatically sent to Obsidian (and therefore DT). It is the only software where I use tags, which either call up the project in use (same folder name as DT and Bookends) or generic areas (#productivity for example).

I’m pretty satisfied, even if I still have some frictions, especially for Obsidian: for example, deleting files often creates missing files in DEVONthink that is very laborious to find and remove, plus the citation system is not so solid.


That is nice workflow. Here is a small trick to find and remove all the missing files in DT in one sweep (I don’t know where I get it, but works great):

  1. Window–>log—>clear
  2. File–>Verify and Repair -->ignore
  3. Window -->log -->select all → Delete
  4. Trash -->Empty trash

First, go to Window and clear your log. Then, go to the File menu, and run verify and repair. Once the process finishes, you will ignore it. That will generate a log. Then, you go to your log (Windows menu) once more and select all the files there and delete them. When you empty your trash, they are all gone.


It’s working. Thanks!

Ideally I’d like to be able to create a smart rule or some macro to automate the process, but even then it’s very convenient.

How do you retrieve the infos after some time, without links and related, if you rely only on search and tags?

I mean: with tags you have the context, but not the “topography”.
When I create a note, I insert text, links (and backlinks), images, all following a logic order, a sort of discourse.
So when I get back to the note in the future I’m still able to recover that logic.

But if all of these resources are stored as separate files, when I tag them I simply group them around a topic, but I lose the logic (the topography).

Are is there something I’m missing?

I ask because organizing notes with links and backlinks is very time consuming.

How do you make your links carry logic (contradictions to the original notion, supporting argument etc, I guess)? Can you give us an example?

Another user of both - I like to take notes in Obsidian because I work (my job) in Windows and can sync between Windows and Mac. Even when working from home it’s handy because I don’t have to walk across the room to other desk if I want to take a note. I have DTTG but don’t like to do data entry on my iPad. If I want to move something from Obsidian to DT I will do it on my Mac. But I use Obsidian for more immediate stuff (daily list, checklists, frequently updated current stuff) and DT for reference material and anything longer term.

1 Like

I’m not sure Obsidian offers much that isn’t already there in Devonthink. Inbound and outbound links, mentions, see also, and the tag cloud feature I didn’t know was there until last week.

@benoit.pointet 's brilliant node map script is the stuff of legend, too.


Interesting use case for people who need to use Windows in certain contexts. Thanks for sharing.

One could index the obsidian vault, then automatically transfer new markdown documents to a DEVONthink database via a smart rule.

This very same use case occurred to me this morning when a windows laptop was introduced into my purview.
In truth it would be overkill for me as I almost always have my phone on me to make quick notes and add reminders but it would be a good way to test out Obsidian to see if theres anything I can get out of it that DevonThink doesn’t get me. That seems more and more like a dwindling target though. DT is an octopus that reaches into many regions of my life.
Maybe theres an argument to be made for a windows based Obsidian system, indexed into DT somehow, that can’t do anything else so it functions as a writing only, and therefore a reduced distraction system.
Always good to test the other options out there, see what everyone is talking about. That part at least is sort of a minor part of my job description, just maybe not specifically to do with PKMs and writing tools.

I use both DT and Obsidian, but this is not revolutionary. I have always used other software packages in conjunction with and alongside DT.

Word and Excel for starters. Until I started using Obsidian, I used Typora as my preferred markdown editor.

Yes, I could do a fair bit of editing in DT, but sometimes an external programme is the only option (i.e. Word) or desirable due to the extra functionality.

I use Obsidian because some of the plugins really jive well with my use case of writing markdown documents. I get the best of both worlds. Yes, I could use DT to do markdown exclusively, but I don’t have to.

I also find the friction of employing my version of Zettlekasten in Obsidian less than DT.

If I could only use one or the other, DT would win hands down. One thing that Obsidian will never do is become a replacement for DT. No way, no how. DT is where I keep all my PARA knowledge, alongside the markdown that is feeding into it from Obsidian.

I do like the Obsidian mobile apps. They are dead fast for reviewing my markdown notes while I am on the go. Yes, I have and use DTTG, but it is slower.


Great video on this topic:


This is the reason why I ditched Obsidian and testing a “DT only” strategy.
After about 2 months I can say I don’t miss Obsidian at all.
Furthermore, I’ve escaped the plugins rabbit hole, and focus more on the output.