I use an Applescript to assist with tag assigment
You mentioned this several times but I haven’t seen an example. I know you must have spent several hours/weeks perfecting your workflow but as someone that got inspired by your use of tags: would you mind sharing some more background to your workflow and how the script helps you? I have very little background in scripting but it seems intriguing to me to have AppleScript assist you in tagging.
True - especially with underpowered LLM models.
But if you use one of the more capable models you can get fairly predictable responses such as:
Among the selected documents find those with the name of a president and add the tag Politics
Among the selected documents find those mentioning electric vehicles and add the tag EV
Among the selected documents find those with the name of a month in the title then apply that month as a tag
Among selected documents find the date each one was published and add that year as a tag
I collect records in the Global Inbox, and process each using my script
The script displays a list of Type- tags for selection
An example is Type-Receipt
If selected, the script displays a list of Vendor- and Budget- tags for selection
and asks for a $ amount
The script then moves the record from the Global Inbox, and assigns the tags
The displayed selection lists are dynamic; built using the tag prefix
2 Likes
Any chance you can share the Applescript?
I am sure with enough time I could figure it out, but having a good reference to get started would be helpful.
Regards.
My dilemma, I have not experimented, used or even know how to get started on using AI, especially for this kind of task.
I’m already lost reading underpowered LLMs and using more capable models.
I don’t know what those are or if they come with a cost or even how to choose a model.
I probably should learn how to use AI, but don’t know where to start, what model to use, what to expect to pay, how do charges accrue, etc.
I appreciate your inputs, I am just lost on how to begin using them.
Regards
Did you read the Getting Started > AI Explained section of the built-in Help and manual ?
We’re off topic for this discussion
I’d recommend starting a new discussion where I’d be happy to answer any specific scripting questions
1 Like
Taking your idea just a step farther…
I think about all the remarkable archives which are discovered decades, centuries later…a trove of an ordinary soldier’s Civil War letters, an unknown Mozart score in a pile of papers in an attic, love letters of a long-gone couple that become modern classics, etc etc, insert your own imaginary example…and the equivalent in our 21st century world?
A trove of 10,000 Day One entries…or a specialized interconnected research graph in Roam (with links out to DEVONthink, Scrivener, Tinderbox…and corresponding unique hooks going the other direction)…or even simply the literary email archive of a modern writer, famous or not, whose thoughts were best expressed among any number of correspondents…?
What becomes of all this material?
In the case of a physical corpus, of course, universities and other institutions built to the purpose take legacy bequests of archival materials, in some cases, and this covers the known stuff…time and chance add to it as may be.
The Wayback Machine attempts to create an everlasting archive of public materials…but what could be an equivalent for private bits? I assume not everyone wants their digital legacy to be “burned” (as Kafka and so many others have directed their executors to do), to at least have the chance of utility or discovery in some not-so-far-off future?
It’s an interesting topic to me, surely someone must be working on it.
Welcome @qraig
Thanks for sharing your thoughts!
1 Like