DEVONthink (not DEVINagent) has been spectacular for me. I have many different documents related to many different projects/purposes, and it does the best out of anything I’ve tried for searching, categorizing, etc.
It has a learning curve. It also seems to be designed with a different mentality than many modern packages (not cloud-focused). I use perhaps 10% of it, and yet still find it with the $. It’s that good at what I use it for.
How does it compare with the FOSS Zotero? Linux users don't get DEVONthink so we need to make do with other resources. I have been trying to pick a reference system for my larger note-taking flow using mochi.cards
DEVONagent is capture. You give it criteria it will use that to decide what searches to run, collect the results, collate map of terms, and present the results for you to peruse. For instance, if you were to search for "Vorlon," you'd get back pages that have the word, but also ranked by other pages found that have the same vocabulary. It will give you a graph of the word and other commonly found words, like "Babylon", "Mimbari", etc. It a good way to explore stuff you don't know because it will show related concepts. It is very niche, but very powerful for what it is for.
Zotero does citations. It lets you track citations to resource sources, add relational information, and organize them. If you have every had to make research papers, having a citation tool very important. When I attended university, it was dirt easy to track and cite supporting material then export it as whatever citation format the teacher wanted. I had one teacher who very strictly wanted a particular version of citation, but it wasn't standard. I made a citation style just for her so I didn't have to remember the rule changes.
It is baffling that there is a video for the $5 version (which is barely more than "do google searches or searches of a pre-canned list of sites, with site previews, from your menu bar" - thanks, I'll keep my $5) but nothing for the $50 software?
devontech has been around for something like two decades and has always struck me as being borderline snake oil. Lots of hand-waving elevator pitch, but I just never see anything explaining how the fuck it's actually useful.
I got DEVONagent for free as part of a package deal way back when I first bought DEVONthink. It’s not game changing, but it’s still quite nice in a few ways:
- You can use advanced query operators like “word1 BEFORE/3 word2” to specify that word2 must appear within a couple of words after word1. Compare with searches that say “oh, you want all documents that contain both word1 and word2? Here’s a dictionary! Here’s your document index! Hope that helped!” (I’m looking at you, Notion.)
- The agent supports that on search engines that don’t natively support that grammar. If it must, it’ll search for all documents that have both “word1” and “word2”, then filter out all the ones that don’t meet the more specific criteria.
- You can automate searches, and get notifications when it finds new results. Not super handy, but still nifty.
- Want to search 23 search engines simultaneously? Fine.
I wouldn’t buy it alone for $50. I’m glad I have it, though, for the rare times that I need it.
There was a point where I had most of my work and ideas in Devonthink. Then I noticed that a few important files had somehow vanished. I don’t know for sure if this was due to the software, but it unnerved me, and I moved to Obsidian.
I used DT primarily as a knowledgebase/notes app; I never found its smart features to be particularly useful. Nowadays, I’d definitely recommend Notion over DT.
I lost four to five years worth of data, the database just vanished. The backups I had were useless (some sort of systemic corruption?). Having been badly burnt, I can’t go back, but I’ve yet to find a tools that matches my workflow as well as DT does. Bummer.
I had something similar happen in Logseq when I stored my main directory in Sync, which is similar to Dropbox. After moving my main directory out of Sync, everything became stable again. Perhaps Dropbox was the root cause of the problem.
The problem could be the way you used the directory. I don't know Sync specifically, but DEVONtech don't recommend storing the databases in directories which will be synced. Rather you are supposed to set up the system to use whatever sync API is available, which presumably makes sure files are internally self-consistent before syncing. It's a while since I set it up, but that's how I did it with WebDAV to my NextCloud server.
It’s an everything bucket. By this point the concept is very well established and has many implementations. Most people don’t care about it, but some do. Snake oil it is not.
> Lots of hand-waving elevator pitch, but I just never see anything explaining how the fuck it's actually useful.
No idea about DEVONagent, but the trial of DEVONthink I've done suggests it would make sense for someone with a few TBs of PDFs (e.g., me). The OCR plus search function makes it like a better Finder. I just can't justify paying $200 for that.
I love DEVONthink, but I understand that it’s not for everyone. What I appreciate is that I can automate an awful lot of routine tasks. For instance, if I drop a PDF of my bank statement that I downloaded from the bank’s website, DEVONthink will recognize that it’s from my bank based on the URL metadata that macOS attaches to it. Then it OCRs my statement (no idea why or how my bank manages to send it as an image), finds the string like “Closing date: X/Y/Z”, renames the file like “Bank statement - 20ZZ-0Y-0Z.pdf”, and moves it to the directory where I keep them all.
I pair it with Hazel, which monitors files that show up in my ~/Downloads folder and does stuff with them. (Examples: moving bank statements to the DEVONthink inbox folder. Opening any *.ics files with Calendar.app so that they’re added to my calendar, then deleting them.)
I could totally do all those things myself. I’d sure hate to go back to having to, though.
> Then it OCRs my statement (no idea why or how my bank manages to send it as an image), finds the string like “Closing date: X/Y/Z”, renames the file like “Bank statement - 20ZZ-0Y-0Z.pdf”, and moves it to the directory where I keep them all.
If you can have it do that automatically... then, okay, I think you've convinced me to do another trial. That's pretty cool.
Oops, my bad. Hazel does the OCR, because the day I was setting that pipeline up, I was able to get it working first. DT definitely does all the rest. I have smart rules set up in it for all the bank / credit card / loan statements we get, certain types of email I archive, and a bunch of other stuff.
In all cases, my workflow is the same: get a file onto my computer and into DEVONthink’s inbox somehow. For email, that’s using the DT plugin in Mail.app to import the message(s). For downloaded files, it’s using Hazel rules to move them into it.
Again, none of that’s stuff I couldn’t do by hand, but neither is really anything we do with computers, right? I’d rather script it up once and let it handle the toil for me every day.
It's not snake oil, it basically provides customized client-side meta search. For people (especially non-programmers) who have search-heavy workflows, I could see it as being a useful tool.
This is what I would always recommend in all the threads about google getting worse - it certainly gets you different results and respects your boolean modifiers.
I like devon tech, maybe it's expensive for a personal knowledge repo but at least it's a buy-once perpetual license. I bought devonthink about 4 years ago and still use it - mostly i'm impressed with how quickly I can browse my pdf library, and it makes backups easy since the databases are single files.
Reminds me of the Copernic Agent (not their current desktop search product). Are there any other metasearch engines that download the results these days rather than stuff everything in the browser?
It has a learning curve. It also seems to be designed with a different mentality than many modern packages (not cloud-focused). I use perhaps 10% of it, and yet still find it with the $. It’s that good at what I use it for.