The new UI for Assistant is worse in every measurable way

v3max

Hi, what is the status of the work on addressing the feedback provided in this thread regarding the new Kagi Assistant view?

It has been two months since we provided our feedback, and nothing has changed…

For comparison, other competing solutions with smaller teams implement improvements more efficiently.

It is truly highly disappointing that the previous Kagi Assistant view was taken away from us in favor of a new view that is dysfunctional and unfriendly. We were told improvements would be coming "soon," and nothing is happening.

dreifach

Worth adding to this conversation that one of the less technically proficient members of my family plan noted that the new UI is not all that intuitive. They said that the old one was cleaner and neater, and that they would've just preferred tags being replaced with folders and to leave it at that.

opfernix

Dear Kagi Team, the alternative solution is to keep whatever UI you want, and provide assistant API.

opfernix

Kagi ruins the UI for power users (minority) to make it look less complex for the commons (majority) so they stay and pay. It is impossible to satisfy both groups, so do not expect significant improvements.

dreifach

opfernix worth noting that the family plan member in my case is not a power user!

BobSmith

I'm still wondering what is going on regarding this. I often refer to older threads I created and it's (also) genuinely annoying not to know what model/ revision was generating at the time, since older models get retired etc. It seems this information is just discarded.

niu3

BobSmith Same, after all the changes I'm just lost in my own un-navigable sauce. No dates, no dividers, no reasonable peek/preview text, no tags, hard to see which model generated which answer. Just one big mess.

BobSmith

Now there is some new thing,

I guess the goal is for absolutely everything on Earth to look like an app for 10-year-olds, and compactness for professional desktop users doesn't matter. I understand very well that you cannot please everyone.

It would be really really really great if there was some user control over this kind of stuff.

BobSmith

BobSmith Arrrggh. To add insult to injury now the "save to .md" logic has changed.

None of the "Contemplating" / "Searched with Kagi" etc sections are now included for some reason. Just the question and the rendered answer.

What is the purpose of all these changes that potentially break stuff for users?

BobSmith

OK I checked, it seems random what parts of the AI "thinking" gets included in the export: for older AI threads it's "some of it". Anything after Sep 3 it's just omitted completely. I have confirmed it doesn't matter if it's .md or .json export.

It's shocking that nobody from Kagi responds here even though they do keep chugging along making cosmetic changes. (That at least people here seem to hate.) It reminds me of Google.

It's shocking that I now can't export part of the AI research that I paid for without resorting to copypaste. It reminds me of Google, actually this part is worse than Google.

I have just today manually exported everything, like 400+ threads, because I'm fully expecting stuff to be just randomly deleted. (Extremely tedious since there is no bulk export option unlike with.. Google)

The sad truth is that I now don't trust that something like this might not happen. Actually based on the experience with this accursed AI assistant feedback thread.

It's with a heavy heart I sense the beginning enshittification also of Kagi.

I have another machine where the original layout / design was cached as html so i could take a screenshot. Just as a reminder to everyone what was lost.

Sorry for the whining.

« Previous Page