I think I understand Devonthink now (Devonthink for academic writing)


Okay, so I spent about a year with Devonthink. It was fine. The main draw was able to get my files on iOS. Apple released the Files app, and I turned into a Devonthink hater (look at my history. You’ll see the hate).

This year I went back to try and organize my PDFs. I only have 800 pdfs, but the process was daunting enough that I was willing to try DEVONthink again. It was fine, it matched some articles with ones I’m reading and it helped me find articles that fit papers I’m writing. That was enough for me to keep using it. The matches were okay, but pretty much what I could already do since I’m familiar with the kinds of articles I have.

I rediscovered DEVONagent, it’s an advanced web search app from the same team. You can create predefined searches and have them run automatically then feed the html pages of articles to DEVONthink. I figured at some point I’ll have my research assistants look through these folders (that had 800 hits each after 3 nights). It’ll help them narrow down their searches.

Today I put in a draft of a paper I’m working on. I click on the “magic hat” that looks for articles that closely match the text. I forgot I had all these new resources in my database from DEVONagent. It pulled up about 10 articles that were perfect for my article. These were mostly articles I did not have in my database a few days ago. Wow.

It saves me space not having to download the PDFs, and I can focus on only downloading the most relevant articles to the piece I’m currently writing.

One tip, don’t have your work computer do these searches at 2am at night. It’ll look suspicious to your IT department. They shut my ethernet port off. Ooops.


I bought DEVONagent Pro, but never got much use out of it (nor put a lot of effort into it). Your use case sounds great. I’ll have to look into it further. Thanks for posting!


I think approaching it as a replacement for Evernote or finde is the wrong mindset. It’s okay at filing but best if you can think of projects where you can throw in a lot of information that you need help looking through.

Some other uses I’m testing:

Filing student emails and projects. If they ask for a letter of rec I can see all my interactions with them, see their assignments, and jog my memory. If I get a question maybe it would help find my responses to similar emails in the past or even find articles that match that students interest.

Use DEVONagent to search for workflows related to academic work. Index pages for me to review later when bored. I find this interesting . I get the irony in automating my workflow to find workflows.


Interesting use of DAP. I typically have used it to do deep searches on specific terms on the web to see what it finds; it sometimes catches things I’d otherwise completely miss with my usual, manual search strategies.


I really had trouble understanding DevonAgent. It looks like a great tool, but I think I‘m too stupid for it.


I had very little luck with it — see The other Devons...Agent, Sphere


I’m torn with putting my writing into Devonthink as I think if DTP as a research database primarily and writing a separate system. Maybe this is misguided.


Can you say more about your process for student files and projects? How do you get the files in, organize them, and resurface the results? I’m asking process-wise, not necessarily technical specifics.


I have the same issue. I can’t decide. I don’t actually ever write inside DTP but I store all my finished things there, including notes and so on. I am trying to avoid having things spread out in different apps as much as possible at present. I use LaTeX and I keep the composition out of DTP until the final PDF which I then store there.


So are you saying the compelling use is really the search/smart search capability?

I tried DT a while ago but couldn’t get past the rather old UI experience on the Mac.

Alternatively, I prefer to keep my source material in sorted folders so the indexing/content is preserved when I back it up to offline storage or mirror it to Dropbox or other cloud service.

To make that work, I have used FoxTrot Search which is a power search system that has both client and client/server options. You do have to learn a bit about regular expressions, but that comes in handy with many other tools.

Not a perfect solution either, but at least my data (files) are still at the operating system level and not inside a custom database or pointed to with links I have to worry about becoming out of sync, etc.

No worrying about import/export or possible lack of future upgrades or support as the data stays in place


@omarruvalcaba can you tell me a bit more about how you set this up? I have DAP and haven’t gotten the hang of it. I use DTP daily for a respository of research articles, but I haven’t automated anything with it. I would love to have devonagent to follow a couple of websites, and perhaps automatically feed urls to DT. Do you leave devonagent running all the time? I have a laptop and don’t generally do this, but I guess I could. If you could tell us more about your process or point me to some getting started tutorials with Devonagent, I’d love to use this more!


Hi there,

can you elaborate on the automated searches feeding into DTPO folders?




Sorry if I miss the a question and for typos. I wrote this on my iPhone.

I think what makes DT compelling to me is the magic matching. Based on the file you’re looking at, it’ll find documents that share similar text.

Devonagent is another piece of software. Same company separate software. But they can work together. Here’s an intro into the feature in devonagent for automated searches. You create a “set” and then schedule it. I funnel results to DT but they have different options like getting an email summary

I turned off most plugins (search pages). I left google scholar, google pdf, and Added a few university libraries. I also try to use hyper specific terms.

For example

First Gen* student university (home or responsibility or work or struggle or stereotype or barrier or “cultural strength” or “social capital”) immigrant (latin* or hispan* or mexic* or chican* or salvador* or colomb* or guatemal* or peru* or ecuador* or venezuel*)

I have Devonagent download the results to DEVONthink. It store the page as resource (mostly text but if you’re on WiFi it’ll load the page). This takes very little space so not a big deal to keep thousands of reference pages in DT.

So I have thousands of references and all my PDFs in DT. The references update each night.

I then drag my piece of writing into DEVONthink and have it look for files that have similar content. This is great for those times you need to find a reference. Or see if you missed a ref. Or find and article similar to an article you know works for your writing or teaching.

I’m still writing in word or Ulysses, but I’ll pop into DT to find additional references or to find an article on a similar topic. I imagine this would be very helpful for meta analyses. Devonagent seems to have great options for the natural sciences. It’s not limited to academic sources.

150 US is quite a bit. I still don’t see it as an Evernote or note system replacement. I prefer hazel for organizing things like receipts. I would only use it if you could think about how you would benefit from matching some file you’re looking at with hundreds or thousands of files.

Eg. Looking for plagiarism in student papers. Looking for similar case files if you’re a lawyer. You write blogs and need a cite. You’re creating a class curriculum and then have it match with your curriculum with a bunch of info pulled by Devon agent.


I’m surprised your school does not subscribe to a plagiarism detection service like TurnItIn or SafeAssign. I imagine that some departments or schools might be philosophically opposed to it, though.


The longer you store information in DEVONthink, the more useful it becomes.

I’ve been using it for years and agree that the matching is phenomenal.


We use turnitin, just throwing ideas out there. A high school might not offer this for example.


We’ve used copyleaks in the past. It’s free (or was the ast time we used it).


I’m starting to do this as well now. I like to compose my writing in Scrievener or Markdown depending on the length of the piece. I’ll slowly then move completed stuff to DEVONThink though this reminds me I need to clean up my mess of files too…


Thanks for that. I won’t use that way but it has encouraged me to use the matching capabilites more than I do and think long term.


thanks for that. will try to use devonagent pro for my medical research in the future. do the result pages contain all the abstract info then? will have to play around with it a bit! cheers