Create individual card summaries for bulk bookmark imports
n
nicole
When you paste a single URL directly into a card, the summary is automatically generated and saved into the notebook immediately.
However, when you import multiple URLs in bulk, summaries are not automatically generated or added to the notebook.
Instead, you must open each imported card, generate the summary manually, and then click "add to notebook" to save it.
This process can be time-consuming and reduces the efficiency benefits of bulk importing.
A setting should be available to automatically generate summaries for imported content and add them to the notebook over time.
This would allow users to import large batches of URLs and have summaries created and saved without manual intervention.
Such a feature would make bulk import truly valuable and save significant time compared to adding URLs one by one.
Log In
t
twan
Added this comment to a similar request:
I had another proposal solution to this problem where some kind of background que stores all requests to recall. It only logs the URL shared and the settings used, then the upload time is nearly zero because of low info intake. Then the que starts working on whatever is logged first without the user having to wait for recall to do its thing.
This way faster UX on the frontend and the backend can work at whatever speed it can. Having a backend sorting system you could acces (like a YT playlist) you could order the incomming logs to whatever your prefer to be done first.
I made this suggestion because of the problem with bulk importing bookmarks or playlists (not yet available). When doing this Recall does not make a summary or connections due to the high cost of importing hundreds of files at once. Which is a viable problem.
However every user has a set limit to upload per month and when reached you have to wait untill the next month. With this que they could effectively just let it run until you run out of credits and resume when the next batch arrives. This is also a perfect way to insert an upsell proposition where you can buy more credits or take a higher tier acount to get more credits for researchers ea.
I mailed this proposal straight to Recall last week, but thought I'd mention it on this platform as well :)
Sankari Nair
Hi Nicole! Right now, we don't do bulk summaries or bulk connections simply because of the AI costs incurred. This might be something that we look at at a different price tier.
b
br.wims
Sankari Nair Why could it not be possible to simply throttle the process so that it is equivalent to a live person adding a url, clicking the summary button, then adding to notebook as though done in real time? The point of the request is to save the mind-boggling tedium of adding them one at a time by hand, or bulk importing (which you allow already) and then performing the rest of the clicks by hand. Putting it up a price tier isn't the answer, at least for me. Simply schedule a job adds the summaries slowly.
As it is, I am being discouraged from subscribing to the first tier, because the free level doesn't give me enough cards to work out whether it is even useful to me. I would suggest an import minimum of 25. That would enable a decent range of topics, and a small enough number to do all the clicks by hand.
Really without some level of automation for adding summaries, I think this is a loser.
S
Sankari Nair
br.wims hi there, the issue isn't throttling or speed. It's the actual cost of creating summaries for over a thousand cards. That's more the bottleneck. We might explore a different tier where the cost is more controlled. But appreciate your feedback and the fact that this is a high demand request.
b
br.wims
Sankari Nair Hello again. I'm probably approaching this from the wrong angle. I have a large number of bookmarks which I would like to have short summaries of content in order to be able to find articles of interest in particular subjects. I have them in the form: url, title . I'm not sure how much work is involved generating a summary from the contents of the site, but perhaps it would be less if instead you used tag categories for each site such as IAB provide, and a quick analysis of the page title could generate its own relevant tags on import, rather than opening the site.
I believe that providing those minimal tags on import would not be a huge overhead, but would provide the kind of foundation for some useful queries where summaries could be generated on an as-needed basis.
--------------------------------------------------------------------------
BTW, if you are not familiar with IAB classification, I just went to Klazify.com and used their demo page to classify getrecall.ia, which generated the following json in less than a second. (If it's inaccurate, you may want to contact the IAB! )
{
"domain": {
"categories": [
{
"confidence": 0.95,
"name": "/Computers & Electronics/Software",
"IAB-602-599": "Software and Applications/Technology & Computing/Computing/Computer Software and Applications"
}
],
"domain_url": "http://getrecall.ai",
},
"success": true
}