Why DEVONthink

Hi All,
I am new to using DEVONthink and have downloaded the DTP trial version to see how well it works for me. I am new to using a document management program and have one very basic question. Why use DEVONthink. The big advantage of DEVONthink seems to be groups and tags. We now have both of these things in Mac OS X. To me it looks like OS X folders are basically groups and tags came along with Mavericks.

What am I missing? What does DEVONthink have that makes so many of you use it so much?

Good questions. Groups resemble folders. But the power of DTPO groups (to me) is the concept of replicants. In some sense, Finder aliases do something comparable. But in DTPO all replicants are equal. In Finder, there is always one original and aliases. You need to treat them differently. That can make a big difference. In DT I have tools to see all replicants. In Finder, I can only find the original from an alias. But neither the original nor aliases can point me towards the other aliases; that’s kind of a dealbreaker. To be honest, if you don’t use filing with replicants, i.e. have only a group hierarchy where each item is exactly in one location, you might not need DTPO (except for using the search, see below).

One of the most powerful features for me in DTPO is searching with the global search window. It seems that Spotlight cannot nearly give me the same level of success. In pdfs, the search terms are highlighted throughout the document. That’s extremely powerful.

On top of that, there are the AI features of DT, which I have not really been able to harness yet, but others do.

I came to DTPO two years ago. I was looking for a system with what one could call hierarchical tagging. I was tired of having to file an item in one specific location. Does a paper go under “relativity” or “atomic” or “spectroscopy”, when it deals with a relativistic effect in an atom, and the experimental technique is spectroscopy? Also, it might have to go into a group “review article” and also “good student intro”. And get a tag “read” and get filed in a group for a given research project I currently work on. DT is ideal for that. Using Finder aliases instead of replicants would be cumbersome, let alone OS X tags, as there is no powerful tag manager in OS X.

In addition, DTPO in conjunction with DTTG on iOS became my de-facto sync mechanism and filesystem on my iPad.

Update: The clipping extension in the webbrowser is extremely useful to me. It will allow me to “bookmark” a webpage in DTPO, but for offline use and instant search, I can save the page content as a pdf (especially useful, a long one-page pdf), which also serves as an extremely reliable archival tool. I can also extremely quickly drag a file or email onto the DT icon or the optional DT sorter to land them in my DT global inbox. That helped me enormously. I sometimes inadvertently find some information while searching for something else, and without having to interrupt my main search, I can instantaneously dump a file into the global inbox, stay focused on my actual work, and then, in a daily close-out, deal with the “treasures” I found during the day (you still need the discipline to do the close-outs, otherwise your inbox will be a gigantic, unstructured pile after a while).

All these things can somehow be imitated with Finder and a bunch of tools, but, at least for me, the workflow would not nearly be as fluent; and that is what counts when we deal with information overload.

DEVONthink integrates the information content of documents of different filetypes much more richly than does Spotlight. It has artificial intelligence algorithms at the kernel of a database and creates a Concordance for each database.

When groups have become populated with document content—especially if groups hold topically distinctive content—the Classify assistant can suggest possible group locations for a new document based on an analysis of the contextual relationships of the document’s text content to the pattern of contextual relationships in the documents contained by one or more groups.

When I’m exploring ideas discussed in documents in a database, I use See Also (or See Related Text for a selection of text) heavily. While viewing a document, when I invoke that command a list of possibly related items is suggested, based on DEVONthink’s analysis of the terms used, their frequencies and the patterns of associations of terms in documents. When I see an interesting See Also suggestion I’ll open it as a new tab in the window, so that I can easily switch back and forth among several documents without losing scrolling position in any of them.

Suppose I see a reference in a document to a single-word term such as hexachlorobenezene. Option-click on that terms and a slide-out drawer appears listing all documents in the database that contain that term.

DEVONthink Pro and Pro Office have large AppleScript dictionaries, so many procedures can be automated and extension of features is possible. Suppose I’m viewing a document that will require action by a certain date. There’s a script for that and it will create a Calendar event and I can set an alert date and time. When the alert goes off, when I open the Calendar event there’s a link in it that, when clicked, will open the document that requires action.

There’s a scripted template called Annotation that I can invoke with a keyboard shortcut, to hold my notes about a selected document. The rich text Annotation note is automatically linked to the referenced document, and there’s a link from the referenced document to its Annotation note. For example, an attorney who has captured the video of a deposition into DEVONthink can create an Annotation note for that video and make notes about testimony, tying each note to the timeline of the video by typing the timeline at which a statement was made. I use such Annotation notes extensively. (I never use Adobe’s primitive plain text, non-searchable notes in PDFs.)

Of course, one might want to create annotation notes about, for example, statements made by someone in several documents. Two of our power users, form and Frederiko, have posted a very powerful script for creating such annotations. See Export an index of annotations to a numbers spreadsheet

These are just a few examples of things that can be done in DEVONthink. It is, as I say, a rich environment for working with information.

All I would say to the OP is the more you use it the more you will discover, and keep on discovering. The best application I have ever bought :smiley:

One would be forgiven for wondering “what gives?” at first…

I distinctly remember having that thought, after I created my first test DB… it didn’t look like much at all, that I couldn’t do with Finder…

Everything changed when I took the plunge, and threw all my data into it… Being able to arrange everything at a whim, was very liberating - especially given how constricting my ‘nested folders inside nested folders’ had become, over in Finder!

Another big call for the use of Replicants. Not sure what you do, but I use it to manage thousands of (mostly) PDFs… Having the option to Replicate – where changes made on the one, are automagically transferred over to the other instance – OR to Duplicate, where identical instances are treated separately, is one of those features that I never realised I needed, and now wouldn’t be able to live without…

The Search functionality is amazing. As is the ability to create an integrated web of pdfs and annotations, through wiki-links.

Lastly, but possibly one of the “features” most easily overlooked - you have this Forum - where DTP old-timers [ 8) ] are prepared to share, support, and update amazing scripts, that start taking DTP into a workflow direction you might never have even imagined… Major, major bonus that!

I have found it possible to get stuff done with the Mac right out of the box, and haven’t felt any particular need to get many apps. I have a lot of data, and I think Spotlight is sufficient. I generally try to keep things simple without contorting myself to fit into the confines of any app.

I started out using DEVONthink a few years ago as a kind of layer on top of my regular workflow – a plus alpha app that produced different results for me when I wanted to find new connections in data or view my data a different way. The “index” feature makes this possible. I’d say I was dabbling in it, though I have built up databases with many tens of thousands of items. In recent weeks / months I’ve gotten a lot more out of the app by contorting myself a bit to take advantage of its unique features like the “See Also and Classify” AI. I expect I will continue to do so, because any effort I put into organization tends to pay off ten-fold. DEVONthink is a rare app that is compelling enough to make me change how I work with my data.

For example, instead of waking up and running a bunch of searches to hunt down stuff for ongoing projects, I just open DEVONthink and everything is already there as PDFs (my preferred format for web clippings). That’s because DEVONagent (a shockingly under-valued and almost unknown app) searches all over the web for me on a regular schedule, collects all of the pages, and places them into my account for me. Remember the opening scene of the Matrix with Neo at his computer and the computer running searches for Morpheus? It’s like that. Except better :slight_smile: With DEVONthink’s AI, sorting it all is a breeze. Then, I sync everything I need for the day to DEVONthink To Go on my iPad. It’s one of only two apps (that I know of) which allows you to securely sync with your iPad without going through the cloud. In fact, DEVONthink has completely won me over with its commitment to our privacy and security with all of its apps.

And, I haven’t even scratched the surface of what it can do. There are so many scripts and so much other advice out there (especially on this forum), you’ll find lots of uses for the app. Over time, it has become one of the most valuable apps on my computer.

As I said above, Spotlight is perfectly adequate, I love to use it, and I could get by just fine with it, but DEVONthink is (for me, at least) a whole new way of working with the computer. This is how I want to interact with my data. Anyhow, you can start out by indexing, as I did, and see where you want to go from there.

For example, instead of waking up and running a bunch of searches to hunt down stuff for ongoing projects, I just open DEVONthink and everything is already there as PDFs (my preferred format for web clippings). That’s because DEVONagent (a shockingly under-valued and almost unknown app) searches all over the web for me on a regular schedule, collects all of the pages, and places them into my account for me. Remember the opening scene of the Matrix with Neo at his computer and the computer running searches for Morpheus? It’s like that. Except better :slight_smile: With DEVONthink’s AI, sorting it all is a breeze. Then, I sync everything I need for the day to DEVONthink To Go on my iPad. It’s one of only two apps (that I know of) which allows you to securely sync with your iPad without going through the cloud. In fact, DEVONthink has completely won me over with its commitment to our privacy and security with all of its apps. /quote]

This is a great thing to be able to do, could you please give instructions on how you set this up? I use DEVONagent but heva never been able to do this :cry: Many thanks.

Here are the basic steps to accomplish the automated search + clipping + save to DEVONthink. There is actually a lot more fun stuff to do with searches as well. DEVONthink has some tutorial videos. The Search Set one is here.

  1. Open DEVONagent
  2. Click on “Window” in the top menu bar
  3. Select “Search Sets”
  4. Choose an existing search set or make one of your own with the “plus” icon on the left
  5. Go to “Actions” in the dialog box
  6. Select “Add Results to DEVONthink” (2:27 on the video)
  7. Select “PDF (One Page)” for the “Format”
  8. Go to “Schedule”
  9. Select the frequency you want the searches to be performed

As pointed out in the posts above, the DT forum is one reason why DT is so powerful. And this thread proves it. Within half a day, the towering figures in DT land (plus me, who is not part of that elite group :slight_smile:) have given you strong answers and leads to find out more.

My first post was somewhat pedestrian, trying to answer your question about “why not the Finder”, which I have asked myself quite a few times in the past.

There is an interesting blog entry, “The case against everything buckets”

al3x.net/2009/01/31/against-eve … ckets.html

Apart from coming across as quite arrogant, I disagree with most of what this guy has to say. But it is a good read to reflect on the subject.

Another thought: While the Finder (and any other file manager on any kind of OS) is very useful and much used by me, it has very limited features and power. That’s classic Apple philosophy. It is even more present in iOS (and ridiculed by the “power user” Android crowd). And I am OK with it. It enables the “masses” to have a nice, reasonably simple system, that is simply good enough. Power users can then add tools such as DT. And also always remember: Apple gives and now increasingly also takes. I used labels extensively in the past. Once they changed their appearance in Mavericks, they became essentially useless to me, as they no longer have the visual power that I used them for in the first place. So I try not to depend too much on Apple-specific features. Pleading with them is useless, so I don’t even try. In DT, for example, I’ve seen numerous instances where the devs responded to requests by single users to reinstate something and add something, within the time frame of the next maintenance release.

Finally a thought on tagging vs. folder structures: I came to tags via iPhoto years ago. There you have no choice. The photos are in a flat database, and you use tags. The way iPhoto and most of such programs are set up, you live in a flat tag space, with nothing but a linear list of many tags. That’s how most tagging on the web is done, and Evernote users do it like this as well.

Well, it bugged me. Example: I tag the photos with my kids with their names to quickly retrieve them. But each time a photo is tagged for one of my kids, I also want the tags “kids”, “family”, “extended family”. So I got into hierarchical tagging. Opponents consider this counterproductive, but I disagree. They would point out that instead of the tag “kids” I can make a smartgroup displaying “kid1 OR kid2”. Fair enough. but this runs quickly out of steam. Suppose I sort photos of animals, and want to display all photos with mammals. I could easily have tags for dozens of mammals (pig, cow, cat, dog …) and not only would a smartgroup be nearly impossible to define, but every time I add a photo with a new kind of mammal, everything would be out of whack.

Hence in my world, hierarchical tags rule. People complain that this is ramming folder structures back into the world of tags. However, I only need hierarchical assignment of tags. E.g. in iPhoto, once the tags have been assigned from a hierarchical list, they live in the flat tag space that iPhoto internally provides, but that’s OK. The hierarchy is only a convenient structure to automatically assign certain tags based on the selection of an associated one.

This worked very well for me, and naturally I wanted to have the same power for my other documents, mostly work-related. That’s when I hit upon DT. Groups became my hierarchical tagging system. 3-pane view or the group/tag window have become the tool to assign multiple hierarchically sorted tags to my documents.

Thanks for this :smiley:

Glad I could help! Let us know how it works for you.

Thank you all for your very helpful comments. I kind of see how DEVONthink may be able to help me get a better handle on my documents but I need to explore the program some more, try some things, and read a lot more. I will be looking through the forums for topics that will help me make better sense of the program.

One thing I do not understand is the practical difference between replicants and aliases. I understand how all replicants are equal but so are all aliases right? It seems that each alias points to the reference document, while replicants point no where. But I don’t see how this difference in any way creates a practical difference in this feature. If I am missing something there please let me know if you have time!

Thanks again!


Hi Steve,

The starting point is what do you want to accomplish? What task did you have in mind when you downloaded the demo of DT? Are you a researcher, academic, administrator, librarian attorney, student, historian, writer or just someone who has come to realise that there must be a better solution to paperless management than dumping documents in a directory. A file system is not a workflow tool, just a more advanced file cabinet.

Devonthink is tough to get into, not because its a difficult tool to use, on the contrary it is superbly well architected, but because a lot of its power comes from you thinking, ‘it would be really useful and save a lot of time if I could do that with my documents’ … I wonder if anyone has tried that before …

These forums have some of the most super helpful support and community I have ever come across for a piece of software. In addition the web is full of useful blog posts on what people use DT for.

Here are a few of my favourites blogposts about Devonthink from completely different fields. They should give you some ideas:


That is a stellar idea ! Thanks and illustrates perfectly what I said above.


Steve, replicants are like aliases, in that they point to a file. But in the Finder, it makes a difference whether the alias is deleted, or the file to which it points is deleted (in the latter case, the alias file becomes a functionless orphan). In DEVONthink, when a document is replicated, there’s not the “original” document and a replicant of that document. Instead, there are two replicants. And in DEVONthink, it makes no difference which instance of those two replicants is deleted. The undeleted instance remains the whole document. (If there are two replicants and one is deleted, the remaining instance becomes a “normal” document; it is no longer marked as a replicant.)

Replicants allow one to file a document into multiple group locations. Each instance adds only a few bits of overhead.

Modifying a replicant also modifies all of its other instances (replicants).

If a document is a replicant, its Info panel displays the number of replicant instances and their locations.

To add to Bill’s comment, I think I am correct in saying that when you index a file it appears in both DEVONthink and on your hard drive. From the user’s perspective, modifying one changes the other, just as you would see with replicants, though there may be a slight lag as indexes are updated.

However, if you delete the file on the disk, then there will still appear to be a “file” with the title in DEVONthink, but when you go to open it, the content will be gone. There are pros and cons to each approach, and that is why they both exist. As I mentioned in an earlier post, I think indexing your existing stuff is a pretty easy way to get started with DEVONthink, because it does not require you to modify your current system in any way. But, you’ll want to be careful about leaving the files in place so that the app can find them.

Personally, I use a mix of indexing, replicants, and duplicates based on my needs. I find indexing especially handy with files that never change. In my case, these are archived files or PDF scans of books.

There are great links above, and Korm mentioned Joe Kissell’s book earlier. It is a nice to have on hand as you come across things you want to explore in more detail such as replicants.
blog.devontechnologies.com/2011/ … e-kissell/

Yep. That is exactly what I meant to say. Thanks for restating it and providing the lovely table. I think that will help people a lot.

Frankly speaking, I think the learning curve is a little steep for DT, but as the table suggests, that is because there is simply so much that it can do. In answer to the OP, DT works for a lot of folks because it is so powerful and flexible. You don’t have to use everything, and you probably won’t, but knowing it is there makes the investment of time, energy, and money worthwhile.

Hmm, I thought I had made that fairly clear in the very first response, and others added to it.

Aliases do only one thing: Point back at the original file. The original does not tell you how many aliases exist, let alone where they are. Aliases know nothing about other aliases. If you delete the original, everything is gone. If you delete an alias, only that one alias is gone. Aliases can be renamed. Therefore, in the Finder it is often not apparent what they point at. You have to double-click on them or inspect with Get Info.

In contrast, replicants are not different objects residing in different places in the file system. There is only one object, the database entry. It has all sorts of flags and tags associated with it (related to group/tag structure, labels etc). It also points at the file. Whatever you do, you always manipulate directly this object.

It should be obvious that this leads to drastic differences in what you can do: Suppose I file a paper. It deals with a relativistic effect in atoms (to take something from my business). I then make replicants in the groups “atom”, “relativity”, “spectroscopy”. In the Finder I would make aliases in folders with equivalent names. So far so good. Now I revisit this item, and I curious how I have filed it. With replicants in DT, I can see all assigned groups. In Finder, I cannot see how many aliases I have made (I could potentially search for the exact filename in Spotlight, but that’s getting quickly unwieldy). So I can’t really judge how I have filed what. You always have to hunt for other instances in non-straightforward ways (for example, I you renamed an alias, you will NEVER be able to find it again, unless you directly hit upon it). Aliases are like one-way streets and you don’t have a city map. Replicants give you the full overview of what is where.

I think you should set up a small database and start working with it. That way you will figure out what this is all about.

From my own experience, I have to say that DT as a program is fairly easy to learn (maybe not all the bells and whistles, but those are not needed right away). What really has a steep learning curve is “how do I find the best organizational principle for my data - and do I have the discipline to maintain it”. That’s truly the hard part, and I still struggle with that.

My opinion is that this is at the bottom of this endless and annoying “Evernote is so much nicer” debate. DT is much more complex and really suited for in-depth knowledge management. We are not talking simple lists of cooking recipes. By the time you put all the tools in place to make DT what it is, the interface by necessity has to look a little intimidating. I bet (but don’t know for sure) that the average Evernote DB is a much, much simpler affair than the average DT DB.

I should have said that the learning curve was steep for me. I really cannot speak for everyone, though I do think (just my opinion) that other people more accustomed to comparatively simple database applications might be overwhelmed.

We aren’t talking about terribly complex concepts here, but they have very different behaviors, even when they sometimes seem to be accomplishing the same thing, and they can have different implications for your workflow down the line. Other information management apps sometimes avoid this dilemma of choice by stripping out functionality, and while this saves you from having to make choices, it also limits what you can do. There is a balance to be struck, of course. I am glad to see that DT has chosen to include so much functionality / customizability. The time put into learning the app will pay off over the years, in my opinion.

I have very few suggestions for improvement at the moment (besides getting DTTG working better), which is rare for me, because I tend to be quite opinionated, but if there is something I want to do, I can almost always accomplish it somehow with the app right now. That’s kind of amazing.