I don’t like the quick answers on search engines, it’s neither fish nor fowl. If I want an AI response I’ll use an AI. So I was using Kagi when I was looking for pages, not synthesising search results on my behalf.
And yeah the AI Assistant came with the free trial and I tried it out, but Perplexity Pro is light years ahead as an AI assisted search, it’s not even close. Honestly it looked like a proof of concept, so I didn’t really factor that into my consideration as I assumed it was still in development.
My experience of Kagi was (I tried it, even paid for it for some time) - Kagi is made for people who are, in some mysterious (or rather I should say “specific”?) ways, predisposed to like Kagi.
For anyone else Kagi makes very close to zero sense (I am one of those people to whom Kagi just doesn’t make sense).
For the former - it might be some kind of a placebo. Or an oddly satisfying feeling of getting to pay for something and form a usefulness bond or so. Or it might actually be useful which we would possibly never know
Since I‘m recommending Kagi to other people, could you explain what about Kagi doesn’t make sense to you? (No worries: I won’t try to explain it to you. I‘m just genuinely curious to understand the perspectives of others.)
After that show, I decided to give Kagi a try. I pay the $10/month plan and it is enough. I compared it against Google Search and also Perplexity AI.
Reasons I chose Kagi:
CEO is a decent human being, who clearly puts the customer/user first. He even promises to find a small way to pay creators if ever makes enough money to do so. Compare that to the Perplexity CEO, he is clear, your data is the product he is selling…
Privacy - my search history is not shared with anyone else
Better search results - because it does things like hide listicles. In google, I find the top 10-20 results are basically the same thing over.
Better Summarization - the Kagi summarization tells me what is sourcing for a claim. I almiost always click through to read carefully. With Google it is blind faith in the AI summary. (I think the fancy name for this feature is Quick Answer. hat tip to @timstringer for reminding me of this)
I’m continuing to enjoy Kagi Search six weeks into my one-year subscription for the reasons you mentioned. I also appreciate features that are unique to Kagi (e.g. being able to rank sites) and find the Quick Answer feature handy.
I’m looking forward to seeing how Kagi Search and the team’s other products evolve.
I’ve been using Kagi for a couple of months now and i love it.
Better search results - awesome
Quick Answer - is surprisingly good
I didn’t realize how handy quick answer would. For years I’ve formulated questions and then rewritten them to make them good google searches. Now I just write out the question, put a ‘?’ on the end and Kagi spits out an answer, with links to the source. This is a huge time saver, because for simple “how do I X?” questions that don’t require weighing options or opinions, I get the answer I need in seconds.
People may be paying for Kagi in protest and disgust toward the amount of crap that comes with Google results.
Take the example given in the screenshots provided in the original post. Assume that the query “Best practices for AI in education” has special significance to you. Assume that you work in education or have a special interest in the field for some reason. Roughly 90% of the screen (on what appears to be an iPad) is taken up by what I’d call nontraditional result blocks.
If I have a vested interest in the relationship between AI and education, perhaps I am an educational administrator, how much do I want to rely on an AI overview for this information? Does the summary improve my understanding of the issue or give me an informed plan of action for how to address it? Does it spark any sort of insight relative to my experiences? Does the minor block that appears next to the AI summary provide valuable links? Would a “listicle" improve my understanding of the issue or give me an informed plan of action for how to address it? Would it spark any sort of insight relative to my experiences?
The next results block. I’m assuming it’s one of those where you click/tap the arrow and a highlighted citation drops down? But it’s not a link to a web page. Which is why I use a search engine—I want links to websites so that I can distinguish the best ones according to the details offered on the page and then read the web pages I select.
The primary role of a search engine should be to index websites and output a list of the pages relevant to my queries. The Kagi results like the ones shown in the OP provide just that, with minimal deviations. The results themselves may not stand out when compared to other services but I think that it should be taken into account that Kagi appears to display more results per page than services like Google and DDG. Probably because they afford less space to things that aren’t traditional links to web pages (like shiny AI doodads and other suggestions and stuff). I’ve also found that there are more eclectic results towards the bottom of the first page on Kagi, results that you likely wouldn’t see at all on the first page of a different service.
Kagi is worth the money if this is sort of stuff matters to you. I imagine that the people who advocate for it the most do not feel compelled to do so in order to justify the cost, but because they may be…undiagnosed information science wonks?
The fact that you have to pay for it is more so an indictment against its competitors. That the results seem to be as good as their leading competitor minus the rubbish and you can customize the arrangement of results (ostensibly differentiating the results according to your preferences) plays in Kagi’s favor.
I’m starting to figure that the discriminating factor that leads one person to prefer a service like Kagi over alternatives like Google or DDG may boil down to that abstract thing called taste. Yes. I think it boils down to that. And it makes sense in a way because it’s sort of what we expect a search engine to align with; the information that the search engine returns should correspond to the discrete interests embedded in our queries. I’m sure this probably sounds like a bunch of hip vapid nonsense. Bear with me. There are people who get paid a minor wage to help figure these things for the benefit of search engine companies.
So my estimate is that Kagi is popular because it does the little things right to such a degree that few people can tell and if you at all attempt to make sense of it you come across as a shill or something because no one ought to think this much about search engines unless they get paid to do so even though they fill such a vital role in the modern world that they may be the first non-human casualty of AI and soon nobody will actually read a web page or select a web page to read themselves for that matter and I do not believe that this is how Al Gore envisioned the internet!
My first internet searches were probably done using Archie or Gopher. And I doubt more than 5% of the wealth of data that is known about me was discovered solely from my online searches. YMMV.
Sadly, I also see this as a tactic for AI Companies to gather even more information.
I stopped using Chrome for 2 reasons. 1) it was a resource hog on macOS, and 2) it was sending all sorts of information back to Google for their search business and I didn’t like that being the case.
If OpenAI provide a “free” browser (which I believe they either do, or intend to do), what’s to stop them using all of the data I access via the browser and also all of the information I enter into the browser to train their models. The same for any other AI company.
That’s a reasonable assumption. But the steps necessary to maintain a modicum of privacy start with abandoning e-commerce and the use of credit cards. IMO. And I’m not going to do that.
I try to protect what I can, usernames and passwords, etc. And live my life knowing it’s the Amazons, and grocery stores, financial institutions, and healthcare professionals, that knows the most about me. And the people they deal with continue to fail to protect my data.
I’m using Kagi for a year now. I was very sceptical first because… why pay for something that is free elsewhere.
But Google got so bad over the last 2-3 years that it’s unusable, at least for me.
I’m a teacher and I usually spend about 1 hour per day looking for new ideas, worksheets, art, doodles and sketchnotes,…
With Google you get 2 pages of Pinterest and Shutterstock, half a page of AI slop, sprinkled with ads all over the place.
Kagi is like Google was at the beginning when it still worked. I find what I’m looking for. It’s saving me so much time…
The number of users is growing steadily, yes, and I hope that will be enough to sustain it. But in times of exponential or even superexponential developments, both the growth rate and the sbsolute number of users is not very impressive.
I am still subscribed to Kagi and use it daily, but not as much as a year ago. Perplexity and Gemini Deep Research are just too good to ignore. I hope Kagi will survive 2026.
I use Kagi by default first because it uses fewer resources. If I can my answer that way excellent.
Listening to the SmarterX AI podcast (can’t recall the proper name), they think perplexity is doomed. They’re involved in enough copyright lawsuits and have no leverage.
