I’m in the process of reworking my DEVONthink setup and wanted a quick sense check.
My plan is to have all my files in Finder (Dropbox), spend some proper time getting a solid folder structure in place (broad categories, minimal subfolders), and then have DEVONthink index everything rather than import.
The idea is that once the structure is right, I can use something like Claude CoWork to help with ongoing filing.. tax docs receipts , scanned notes etc , as that’s the bit I tend to let slip over time.
Inside DEVONthink, I’m keeping things very simple:
Folders are organised by areas of life (home, music career, etc.)
Project folders live within those areas, not as a separate system
I use just few tags for showing current projects:
project:…
status:active / backburner / completed
So for example:
Home area/Garden Studio Build/
The project lives there permanently. While it’s active, I tag the project folder so I can surface it in Smart Groups. When it’s finished, I just remove the tags — nothing needs to be moved around when the project is completed
So effectively:
folders = where things live
tags = what’s currently active
my main question here is:
are there any pitfalls to having your whole files live outside of DTP4 and have the whole indexed like this, and relying on a small number of tags just to surface active work?
Would be great to hear if anyone is working in a similar way or sees any issues with this approach.
Remember that Finder is not a “container” but is view into the macOS file system with certain features to deal with those files, folders, and remote network resources. DEVONthink is not a Finder replacement. Indexing is different than Importing and brings its own issues and risks which are not only discussed often here on this forum by many (search) but fully explained in the DEVONthink Manual. If not understood could lead to “surprises”.
Me, I rely fully on DEVONthink’s features and power and don’t delegate to any other source of “intelligence”. Just me.
You have two objectives. One is to re-organize your structure of folders and files at the Finder level. The other is to apply tags to indicate various needs.
My question was more about Claude CoWork. Personally, I’d not put any sensitive information on any US-based cloud service. Much less let a US-based AI read and analyze them.
Yes I see what you mean now. Claudework would be pointed at a folder on my computer. That folder does exist on dropbox as well. As far as I know Claude Cowork is working locally but I must check that
Because manually tagging in DT does nothing extra that manually tagging at the Finder cannot do. Why add extra friction to have to open DT to do what can already be done when working directly within the Finder?
DEVONthink already has its own “Classify” function. Have you tried it?
What about smart rules? Automated filing is a common use for them. I have no experience with Claude so I don’t know how they compare in practice. I’m personally not interested in using LLMs.
There is even a Classify Via Chat action for smart rules and batch processing in DT4. I assume this is more private and secure than Claude Cowork, so I would give that a try first.
To answer your main question, I’ll tell you the same thing I’ve told others here before:
Many years ago I used to index everything in one particular database because it was necessary. I never ran into any big issues like some of the horror stories here. Little issues here and there with files gone missing or whatever else because of something I did, but nothing catastrophic.
That said, I was always mindful of all the caveats and warnings that have been posted here repeatedly over the years.
So, there is no issue doing it if that’s what you need. But you need to be mindful that your setup is more fragile than if imported, and therefore be prepared if you inadvertently “break” something.
I am also now going back to indexing some things particularly for use with Cowork/Code.
What @rkaplan said. Dropbox is designed to propagate mistakes, not protect against them. True offsite backup tools exist, but Dropbox isn’t one of them.
are there any pitfalls to having your whole files live outside of DTP4 and have the whole indexed like this, and relying on a small number of tags just to surface active work?