Tinderbox Forum

Tinderbox notes on Mac for DevonThink indexing?

I have very recently been nudged towards using DevonThink for catching ideas across platforms and realized when testing it how powerful it was when indexing my almost 3000 pdfs in Bookends to find overlooked links between them (impressive AI). I am now wondering if it could be useful for also finding potential links between notes in large Tinderbox documents (a la Zettelkästen). Here comes my two questions:

  1. Are the Tinderbox notes stored in a folder somewhere to be indexed directly in DevonThink? I feel really dumb not knowing where my important notes are anyway…

  2. Would using DevonThink to index the folder with the notes in Tinderbox cause any problems for Tinderbox? There are no problems for Bookends.

I am very grateful for all the kind assistance I get on this excellent user forum. I hope one day to learn enough to also be able to contribute answers, not only questions.

Best wishes,

Tinderbox files are XML with the one file containing everything in your TBX document. Great for long term availability of your notes in a non-proprietary format.

Indexing the TBX file in DTP however will not be useful.

Linking between DTP items and TBX notes is fairly well supported and may be a more interesting route for you I think. Try dragging an item from DTP into your TBX window and see what happens for example.

Worth browsing the forum for DTP related topics.

No; Tinderbox notes aren’t stored in separate files. But you might want to look at Tinderbox’s own “Similar Notes” list (⌘-7 in the text pane)

Thanks for suggesting better (and simpler) ways of getting a workflow between Tinderbox and Devonthink. I’ll give it a go and see what happens.

It looks like “Similar to” in Tinderbox is just what I am after, and it seems like it is integrated in the “Links Pane”, which I have always open anyway. Great! I guess I am trying to figure out ways to use Devonthink Pro, which I think I will buy a full licence for just to get its AI to help me navigate my ever-growing Bookends library. What is Tinderbox looking at when suggesting notes to link to? Only text content and attributes, as in Devonthink, or is it also looking at already existing links?

Merry Christmas Mark, and thanks for an amazing software that is just getting better and better! For instance, the “Dance” functionality is amazing to order to automatically provide visual structure in large documents without hierarchy (e.g. Zettelkästen).

Hi @Per, BTW, you may find this note useful.

The conversation of Tinderbox vs. Devonthink is not a this or that one but rather a this and that one. They both do what they do really well and there are ways for them to work together.

A few of us a few weeks ago revisited the conversation of how Tinderbox and Devonthink can work together, Revisiting Devonthink Annotations to Tinderbox. Specifically, we were looking to solve the challenge of how to get Devonthink annotation easily into Tinderbox for analysis. I started a process, a hack if you will, that got the conversation going again and then @Pete swooped in and saved the day with an AppleScript and method that lets you easily get your annotations out of Devonthink and into Tinderbox.

Let us know if you have any questions.

I use both Bookends and DEVONthink (and have done for over a decade) and it might be worth pointing out that the relationship between them can be slightly uneasy, depending on how you do things. To judge from posts I have read on various forums, a fairly common way to approach things is to let Bookends look after pdfs, while keeping them in a specific attachment folder, and have DEVONthink index that folder. This means that you can use DEVONthink’s advanced search functions, but you cannot move attachments into other folders, otherwise you will break the link with the entry in Bookends (you can replicate the attachments, but it gets to be a bit of a faff, in my experience). It is worth looking around the DEVONthink forums, where there is some discussion of this sort of thing, and some scripts that might be useful (e.g. https://discourse.devontechnologies.com/t/linking-bookends-reference-to-devonthink/48208).

Lurking behind this is the question of whether it is best to use the so-called AI of a program to help with organising material, or use the genuine intelligence of the human brain, which (often) does a better job, though with the cost of greater effort. It might be advisable to look on the forums at DEVONthink to gain an idea of how the automatic categorisation of material works (if I remember correctly, it basically depends on looking at the words in the main text of the item – so if you have a concept that is not specifically mentioned in the text, it does not affect the result).

Hi @satikusala,
Thanks for this! I will have a look. However, when it comes to annotations in pdfs stored in Bookends but indexed, analysed and sometimes annotated in Devonthink, it is very straightforward to get them into Tinderbox through the Tinderbox-Bookends integration.

  1. All notes and highlights done in Devonthink follow the pdf-file to Bookends
  2. Link Bookends to Tinderbox (in Bookends: File/Link to/Tinderbox)
  3. Select the notecards in Bookends that make up the basis for a potential note in Tinderbox and send them with their citations and page numbers to Tinderbox (in Bookends: Copy Notecard and Citation): Tinderbox window opens automatically and the new note is selected

This works really well for my scientific work flow. If I learn additional ways to use Devonthink and want further integration with Tinderbox, I know where to look for input. Thanks again!


Thanks for the heads-up!

Cool! They’re having a winterfest sales, see https://www.artisanalsoftwarefestival.com/. If you buy now you can save 25%.


Why not both?

I find DT’s AI invaluable when I need to categorize a large number of files at once, for instance if I’ve just imported the entire proceedings of a conference. And the “See Also” feature is useful for looking into corners of the database that I’d forgotten existed. My main database is well over a million words and spans more than a decade; it’s easy to lose track of what’s there.

But the AI works better if I treat it as an assistant, not an alternative. That is, I’ll use the first dozen or so papers in a conference to show it what the categories “should” be. I use “See Also” suggestions as indicators of spots where the structure needs fine tuning or consolidation, and so on.


I wasn’t trying to suggest that they are mutually exclusive. The point I was trying to make was that sometimes the AI cannot do the work for you. But a lot may depend on the kind of work you need to do. It seems like our kinds of work are very different. Before I retired, the two major projects I worked on (by myself, not as part of a team, I ought to add) took eight years and five years from beginning to end, resulting in two pieces of written work. I have had DEVONthink databases comfortably exceeding four million words of text. (Actually, I’ve just looked at my main database and I’m surprised to find that it has over 20 million total words in it.) But in that period of over a decade I might have used “See Also” once every six months. In the first project that was probably because the arrangement of the material had to be chronological, and had very little to do with what was actually in the text. Then again, the material I was collecting was in four different languages, which might pose a challenge to the AI. In the second project, there was a mix of chronological and thematic. The themes only emerged as a result of reading, analysing, coding, and interpreting the text. Tagging became quite useful in that case. In fact, tagging is probably the feature I have used most in DEVONthink, and I also made a certain amount of use of replication, as I found items that belonged in more than one location. Same tool, but different approaches, I guess. And different fields of work.

Buon Natale!

Edit: too much to drink with lunch – I can’t have been using DEVONthink on the first project because it was too long ago, but the data from it went into DEVONthink later. But the point about the way the material was organised still holds. It was chronological rather than based on the content. There were some interesting posts about using DT in historical research posted many years ago here: https://idlethink.wordpress.com/2011/06/24/on-devonthink-and-history-research-i/

1 Like

@eastgate and @satikusala , just to follow-up:

Tinderbox’s suggested notes in the links pane is really good at indicating potential links between notes in my “Zettelkästen on steroids” (Tinderbox is truly powerful for this type of knowledge management system), so I am happy with that without involving DEVONthink. However, it is also useful for me to get suggestions on what publications in my rather large and continuously expanding Bookends library that may be related to a particular Tinderbox note. This is very straightforward though, thanks to the three softwares seemingly speaking the same language. I simply copy the note in Tinderbox and import in into DEVONthink (Data/New/With clipboard), which keep an index of my Bookends library. Voilà! DEVONthink’s “see also”-function provides a list of the scientific publications (with the strongest semantic links to the note) to start with. This is really great and would in my mind not benefit from automation, since it concerns a particular note and takes literally just a few seconds to perform.

Getting DEVONthink notes and Bookend notecards back into Tinderbox as input for new notes/zettles is very easy thanks to Tinderbox’s “watch groups from DEVONthink” (for capturing ideas freely on my Mac or iPhone) and the process described in this thread (for capturing annotations in the publications with citations and their page numbers). Simply brilliant!

1 Like