[HankWozHere]

Kagi Assistant allows you to do search with LLM queries. So far I feel it bears reliable results. For instance - I tried couple of queries for product suggestions and came back with some good results. Whilst it’s a premium service , I find the offering to be of good value.

[chrisweekly]

Yeah, Kagi's search results are so much better than Google's, it defies comparison.

[rglover]

Just switched my default to Kagi based on this comment and you're right. It honestly feels like old-school Google before all of the algo changes.

[abtinf]

You were paying for Kagi but not using it as the default?

[rglover]

No, I only noodled with the free/trial searches before, but this reminded me to pay and make it my default.

[eli]

It's neat but I've found the value kinda variable. It seems heavily influenced by whatever the first few hits are for a query based on your question, so if it's the kind of question that can be answered with a simple search it works well. But of course those are the kinds of questions where you need it the least.

I find myself much more often using their "Quick Answer" feature, which shows a brief LLM answer above the results themselves. Makes it easier to see where it's getting things from and whether I need to try the question a different way.

[wongarsu]

The quick answer (ending searches in a question mark) also seems pretty resilient to hallucinations. It prefers telling you that something wasn't mentioned in the top search results over just making something up

[szszrk]

There is one more aspect of Kagi assistant that I don't see discussed here. I'd love to support some "mass tipping jar" service and/or "self hosted agent" that would benefit site owners after my AI actions spammed them.

You can simply just pass it a direct link to some data, if you feel it's more appropriate. It works amazingly well in their multistep Ki model.

It's capable of creating code that does analysis I asked for with moderate amount of issues (mostly things like it used the wrong file extracted from .zip, but it's math/code is in general correct). Scraps url/downloads files/unarchives/analyses content/creates code to produce result I asked/runs that code.

This is the first time I really see AI helping me do tasks I would otherwise not attempt due to lack of experience or time.

[dmazin]

Has anyone compared Perplexity with Kagi Assistant?

I am always looking for Perplexity alternatives. I already pay for Kagi and would be happy to upgrade to the ultimate plan if it truly can replace Perplexity.

[Zambyte]

I had been paying for both for several months, and I decided to cancel Perplexity about a month ago. First and foremost, I feel like the goals of Kagi align more with my goals. Perplexity is not afraid of ads and nagware (their discover feed was like 30% nags to turn on notifications at one point if you had them disabled, and it's still an annoying amount). I also really like the custom assistants in Kagi. I made a GNU Guix lens that limits my search results to resource related to Guix (official docs, mailing list and IRC archives, etc.) which I can access with !guix, and I made an assistant that uses that lens for web results that I can access with !guixc. I can ask something like "how do I install nginx?" and the answer will be about Guix. You can do some customization with your bio on Perplexity, but it kind of sucks tbh. It would randomly inject info about me into completely unrelated queries, and not inject the info when I wanted it to.

[kedarkhand]

Would you be willing to share how did you that? New to both Kagi and Guix!

[Zambyte]

I can actually just share the lens I made directly with you!

https://kagi.com/lenses/l7mPOuJp7zljHquBjsekFn6dM9Thw1A8

I'm not sure if adding that to your account will include the configuration I have set to access the lens with !guix, but if it does not, you might want to add it. The lens basically just uses this pattern for search result sources:

logs.guix.gnu.org/guix/, lists.gnu.org/archive/html/bug-guix/, lists.gnu.org/archive/html/info-guix/, lists.gnu.org/archive/html/help-guix/, lists.gnu.org/archive/html/guix-devel/*, guix.gnu.org

I don't think I can share the assistant directly, but if you have Kagi Ultimate, you can just go to the Assistant section in the sidebar of the settings page, and add a new assistant. You can set it to have access to web search, and you can specify to use the GNU Guix lens. You can pick any model, but I'm using Deepseek R1, and I set my system prompt to be:

> Always search the web for answers to the users questions. All answers should respond relating to the GNU Guix package manager and the GNU Guix operating system.

and that seems to work well for me. Let me know if you have trouble getting that set up!

[pbronez]

I got a free year of Perplexity thanks to owning an R1. I already had a Kagi subscription, but decided to give Perplexity a try.

I found Perplexity was slower and delivered lower quality results relative to Kagi. After a week of experimenting, I forgot about Perplexity until they charged my $200 to renew my free year. I promptly cancelled the heck out of it and secured a refund.

[hooli_gan]

Does it just start a search or does the chat continue with the results? Would be cool to continue the chat with result, which were filtered acording to the blacklist.

[lemming]

The chat continues with the results, and I often explicitly tell it "search to make sure your answer is correct" if I see it making stuff up without actually searching. I use it multiple times a day for all sorts of things.

[KoolKat23]

I have a subscription, please could I ask how you do this? I only know of the append ? Feature.

[arandomusername]

You need their ultimate plan

https://kagi.com/assistant

[KoolKat23]

oh okay, thanks! (too rich for my blood haha).

[enterpriss]

https://kagi.com/assistant

[KoolKat23]

Thanks