Using DEVONagent to search and summarize papers in DEVONthink database


In DEVONagent, the digest view with the map is great way to study directly from a search without having to open individual documents. Can DEVONagent perform search and summarize on a DEVONthink database? I have stacks of research articles that I want DEVONagent to summarize but I can’t figure it out. Thanks!


This is ony possible after enabling the web server of DEVONthink Pro Office 2.x or DEVONthink Server 3.0 and using the plugin “DEVONthink Server”.

1 Like

Is this what the default user “DEVONagent” in DT3 Server is used for?


Exactly. You can now specify which databases can be searched via DEVONagent without having to make them public for everyone else.


Thanks for the help, it’s working great now!


This sounds interesting! I’m trying to test this feature, but I’ve registered DT3 with my license for Pro only which doesn’t include the Server functionality. While in beta, I should be able to also test the DEVONthink Server though, right? Would I just remove my license temporarily (without complications down the road)? Or should I test it with DTPO 2.x?

And what’s the difference between the DT3 native option of searching, selecting a relevant term from either the list- or cloud view and (a) go through the occurrences in the document and (b) interact with the similar words mind map?
(via DT3)

Am I assuming correctly that DEVONagent would present me with a continuous view of relevant documents, displaying excerpts of all passages where the relevant search term appears – across several documents? I.e. so that I wouldn’t have to go through each document manually?

Sorry for the probably naive question, I haven’t really delved into DEVONagent’s capabilities yet.


Both options are possible.

DEVONagent’s Digest tab includes both a summary and a topics list. Selecting a topic shows a similar map like the one of DEVONthink 3’s Concordance but in case of DEVONagent the similarity is only based on the search results, in DEVONthink 3 on the complete database.

1 Like