Generative AI continues to dominate tech discussions, and with DEVONthink 4 Copernicus, we’ve integrated comprehensive AI support. This raises the question for some people in our forum or elsewhere on the internet: Is DEVONthink now an AI application?
Actually, DEVONthink has always been an application with AI integration. Since version 1.0, DEVONthink has been powered by our own artificial intelligence, a technology that still forms the core foundation of our apps, enabling many features like the powerful search or the contextual See Also inspector. However, this isn’t AI in today’s popular sense, like a chatbot. It’s rather a local machine learning engine that allows DEVONthink to find and use relationships between documents. That’s why DEVONthink’s original application icon featured a brain in a glass ball — perhaps some long-time users remember?
With DEVONthink 4, you can now seamlessly incorporate generative AI into your workflows. Use chat models like Claude, ChatGPT, Mistral AI, and many more for text work, integrate them into automations, or “ask questions” about your documents. You can also use image generation models like OpenAI’s Dall-E or Flux to create visuals (the title image, for example, was created with GPT-Image-1 in DEVONthink).
But, and we want to emphasize this, you don’t have to. Generative AI remains your private matter and is always opt-in. All feature using external AI remain entirely optional, complementing both the familiar features from previous versions and the many new non-AI features we’ve added. And as long as you don’t specify a model in the settings, whether local or server-based, DEVONthink cannot access them. So whether DEVONthink is an AI application is ultimately up to you and your own way of working.
We also know that some of you are not interested in generative AI and might even find the topic tiresome. That’s why, in the coming weeks, we will discuss the many other new features and changes of DEVONthink 4 in a series of blog posts that focus beyond AI (though a mention might slip in). And if you are interested in AI but unsure where to begin: In our post Where to Find More About AI? we show you two good options to get started.
I think the quoted paragraph is particularly interesting, because you clearly feel the need to emphasise that the new ‘Chat’ integration is completely optional. I think you’re absolutely right to do so – generative AI is divisive (increasingly so as it becomes more prominent and the failings become more reported) and many people want nothing whatsoever to do with it, if they can help it. And I’ve read enough comments outside these forums from people who assume that ‘DT4=Gen AI’ to think that the issue is a distraction from DT’s real benefits for some people.
That’s why when the first public beta was released I suggested that it might be a good idea to have an option to hide the Gen AI stuff in the interface – much as you do with the ‘Server’ stuff – rather than the current situation where the features are prominent in the UI, but won’t work unless you specify a model.
The former allows people to feel they have control over whether and when AI is used. At the moment, there are Gen AI menu items / settings all over the place and this adds to the feeling that Gen AI is something that can’t be avoided, even if it can be in DT4.
Of course, this is only a feeling, but Gen AI is highly controversial and feelings matter – unlike the case with other features you may never need: you’ve never felt the need to write a blog saying “Don’t worry, the PDF features are completely optional”, for example!
I’m not suggesting at all that you should have two different versions, of course, but I wonder if there’s a way to let users switch the visibility of the AI features on/off, it would help remove some of the distraction of the Gen AI issue.
I am confused by a comment and one of the post talking about AI and Devonthink.
They said that AI does not talk to Devonthink. Devonthink talks to AI. I’m not quite sure what the differentiation is there.
Several of the productivity Youtubers have set up Notion and uploaded their journals and other information that they have gathered and then use notebook LM to essentially talk to their second brain.
I have uploaded the Devonthink manuals into ChatGPT and it has been really helpful and explaining some of the concepts that have alluded me today.
I have ADHD and I have gathered quite a bit of information on how to optimize my productivity given my special brain chemistry.
I have uploaded some of that information into ChatGPT and used it as an ADHD Coach.
I’m hoping that with Devonthink 4 I could talk to the ADHD PDFs.
I can talk to some of the PDFs using PDFpal that is available on setapp
I sense a lot of trepidation regarding artificial intelligence and the Devonthink forums.
I could be wrong, but if you use Devonthink to create a second Brain, then being able to interact with it, using artificial intelligence could be a great productivity advantage.
I view it as the ability to turn data into knowledge
Is there a video demonstrating the use of AI with Devonthink. ???
If you use, Devonthink has your second brain how can you interact with it in version 4
