I was a skeptic at first. I subscribe to ChatGPT, and use the app on my Mac and iPhone daily. Why do I need that kind of thing built into DEVONthink?
But I am finding it useful. I use it to ask questions about documents, groups and whole databases. Sure, I could do that by uploading documents to ChatGPT or NotebookLLM — but why go to that extra work if the documents already live in DEVONthink?
How are you using DEVONthink’s new AI?
So far, I’ve only been testing functionality and comparisons to using local LLM models as I do via Open-WebUI and Ollama. The fact that DT’s implementation allows use of Ollama and local models was a very welcome surprise when this came out. I’m still in the early phases of exploration, but what I’d like to do is create a workflow (whatever the tool) that allows me to have a collaborative creation of a document where I can continually guide, review, change, and control the AI as the document is iteratively created. I’ve learned from experiments that there is a need to really control that process today to achieve what I want but haven’t yet tried to do so with DT/AI. Looking forward to it though.