8 of the top 10 "raised" sites are software dev sites and with #6 being MDN I'm guessing it's not just any software devs, is web developers specifically that use Kagi.
Am I drawing the wrong conclusion? Does that mean Kagi's days are numbered? What would it take for them to get enough non-web-devs that the top 10 raised looked more representative of the average internet user?
Technical users also likely have specific sites (namely documentation) that they wish to bump in their results where other professions might not.
I'm not disagreeing with your point, just offering some thoughts of my own.
To be honest, I’ve only heard of Kagi on HN and nowhere in any of my friend groups.
Why would it mean that their days are numbered? Nothing wrong with having steady income from a loyal customer-base, even if that customer-base is niche.
Growth mindset is a big part of our sick society, unfortunately. It's the only thing our politicians like to talk about, after all. Being a stable business delivering value is as good as dead.
edit I do not agree with this, but this is what i assume OP was referring to? Because right now web devs is a quite populous and active crowd.
Or how about the average knowledge worker? There's hundreds of millions of them.
I pay for Kagi and it's refreshing to actually see real results rather than sponsored slop from the others.
Loyal Kagi customer here, based on their posts and in my dealings with them, they are doing their thing and doing well
They are focused on privacy, do a great job of it, and their AI assistant is top notch (highly recommend you take a look, can choose from many models and swap out responses instantly, not even getting into the awesome search features)
Not commenting on your (good) who is the main audience question, rather the other point about if Kagi is doing well
I subscribed my girlfriend to it as well and tell people whenever the moment is appropriate
Really rooting for these guys to succeed long term
As an aside, when I got my Kagi subscription the first thing I did was lower Pinterest results
Here are the stats: https://kagi.com/stats
How many are paid vs trial accounts?
We know family doesn't offer a trial nor teams.
126 teams x 5 or 6 members = 1,000 accounts at 10 per day 10k
4500 family plans: most will take the 20 a month plan 100k
45000 individuals lets say they are all paid most on the 5 dollar plans lets assume on average 6.50 is earned 300k
Then you have orion+ members at 2000 giving an extra $15 per account. 30k
They probably make 450k a month
They have 19 employees on linkedin and they are listed at under 50 everywhere else. Lets give them 25 employees at 100k average salary which would be 2.5 million in salaries which might be low.
Add on costs to actually run the website (paid search, servers, office costs) which hopefully cost less than 3.5 million.. the rest is profit.
I'd say they are doing well enough. My average of 5/6 per team might be much higher if they have a few 100+ sized teams. I think the mode would be 5/6 regardless of the average.
It’s also much bigger than their customer base. Keep going, Kagi.
I do both, and I have MDN pinned. The thing is, if I search for C#, Python, Java, I get great results on any search engine. If I search for web stuff, I get tons of crap and mediocrity. MDN is almost always what I want instead of some other stuff.
Regarding "days numbered", I agree with the gist of the other replies ;)
Perhaps it just means technically inclined people are the first to see value and use in some product. Whether word of mouth spreads and people will find value in paying for search engine + goodies - time will tell.
Kagi is fairly good at ranking results and essentially making do with what they have, but in my experience it does not seem very good at all for searching for anything particularly obscure. It's like how DuckDuckGo uses Bing - nearly useless if you're not searching popular sites (like Wikipedia, MDN, etc.)
With Kagi this is more significant because paying for a search engine is an even larger obstacle than simply switching.
It is developers who are more likely to take digital privacy more seriously than the rest. And so, it seems like Kagi is indeed on its way finding product fit with the most demanding segment (from a digital/information privacy standpoint) there is.
For any tech, it is fully expected that early adopters belong to a niche. Whether Kagi wants to make the leap or stay in this segment is upto them. There is no indication that they have saturated the market even in their niche ("developers"), while this niche itself on its own might be lucrative enough.
Besides, Kagi may very well venture into other products to sell to the same market segment.
If anything, to me, the writing that's on the wall for Vlad (Kagi's creator) right now is... "The world's your oyster".
It’s better to “own” 10% a known, faithful customer base than 0.1% of a floating one.
IME, this works well.
If they can sustain, maybe they can takeoff. Search in GenAI world is hard, and Google has other focuses with talent and inference chips too.
I hope their days aren’t numbered!
I think it needs some UI improvements. It’s ugly, and I find it can hinder actual use.
More usability improvements on features. There’s a lot I’m still not leveraging because I haven’t bothered learning. Maybe they can build an LLM tool to help with this?
And I don’t care how, but make it easy to make it default on all browsers, mobile mainly. Maybe they fixed this recently, but when I was swapping browsers recently, this was annoying. If they can’t fix this, they probably won’t make it.
Just some top of mind thoughts as high-usage, 95%+ non-coding user.
The "hacker" type of computer user is zealous in everything they do.
What's important is that Kagi refrains from letting hacker zealots influence their business too much, because normal people also deserve high quality search results, and most knowledge workers are not software developers.
I never understood why Google let them destroy Google Image search results.
I'm firmly a Debian shop, but I find that the Arch Linux wiki often answers non-distro-specific questions that I have about how to do something on Debian.
(Especially since I'm usually using Xmonad without a lot of the "desktop environment" stuff.)
But sometimes it just has the answer you need in an easily digestible format. Top 10 source for me, but not a top 3.
-- Some nerd with almost two decades of distro-hopping experience.
But for getting pointed in the right direction about things that have been obscured by the desktop environments, and then left largely undocumented nowadays, the Arch Linux wiki usually points me in the right direction.
Much of it would be pretty confusing to someone who only wanted high-level documentation in terms of the Gnome Desktop or KDE Plasma, though.
I briefly used Manjaro, Debian and Ubuntu before that, and now am primarily a NixOS user, yet I still find myself coming back to the wiki.
I wish geeksforgeeks would die in a fire, no offense to its operators.
I stopped wearing T-Shirt swags from companies quite a while back. Recently, I thought of promoting Kagi and wore the T-Shirt they sent at a few meet-ups, office spaces with lots of tech-people and no-one recognize it. A few of them thought, when we talked, if the logo is for a Golfing group/community!
Personally, I was thinking I’m proudly promoting something akin to ‘Wearing Google T-Shirt in 1999’ but this time, “Humanize the Web.”
It makes me think that for Kagi customers, search engine rankings optimize for something other than useful sites such as docs.python.org and cppreference.com
Runner up to the NYT
IMO I like the fact that they link sources to their claims, which is very rare on the current web. I think of it as a somewhat trustable source of information. Am I wrong?
If you're on mobile you'll get asked to install the app, if you try to view a video on mobile you'll probably get asked to solve a captcha and again asked to install the app.
Say you search something related to repairing a bmw, you might get results that are "bmw mechanic", "bmw repair inclusive", "bmw maintenance meme" which just lead to those tiktok discover pages.
Making both completely useless as sources
So many games have moved to different wiki sites but because of SEO the Fandom.com wikis still appear near the top of "normal" search results, unfortunately.
The sad thing is we should be living in a golden age for these sorta websites, but Google has to promote dogshit so nothing cool gets built.
Mood boards do nothing for me, but that isn’t true of everyone.
I don’t recall using this list to populate mine but I think I must have!
Is it because it is paywall or for political reasons?
So while those results-gated-behind-logins might seem annoying to us, we are in the minority.
[1] https://www.statista.com/statistics/463353/pinterest-global-...
†: hell son that was during my dating years. I was going on a lot of dates with women specifically in that demographic. Once I got to know a woman I would research her on google while narrowing my search to Pinterest. Out of 20 or so women I recall there were only 2 that didn’t have Pinterest accounts. The rest had, very active, Pinterest accounts.
I remember years back being in some social thing where someone was railing on social media and pontificated "absolutely noone wants what Facebook is", and I grudgingly had to retort "you don't want Facebook, but a half a billion people (at the time) obviously do".
Read the room.
I find art, tattoo ideas, home decor, concrete/wood builds, clothing fits, wedding themes…
It’s like a one-stop-shop when you need some fresh ideas on design.
Unless, as you suggest, they take over Google Images but not text search results? I could believe that I use Image search sufficiently rarely that I wouldn't have seen a Pinterest result.
Compared to the AI slop flooding image search results, pinterest is increasingly looking "better than the alternatives".
People may also just find that they don't get a lot of value from it, but if that were the only reason I'd expect geekforgeeks to rank equally high with it or higher.
It, however, has a long, well-earned reputation from being likely the worst site on the net for a decade, back in the early days. Even if I have a question that I think it would answer well, I still recoil at the thought of clicking a link to it.
For another example, I gather that most of the high school chemistry I learnt is wrong, but it was mostly pretty useful in understanding the physical world as far as I needed to progress in my education at that age. W3Schools can be useful in that respect.
I've used it for looking up SQL, XML/XSD/XPath and such when I had some specific question. Found it quite helpful, and still use it as a quick reference every now and then.
Nowadays with AI slop websites all over the place, plus formerly legitimate newspapers not attributing sources & edits either, this may not seem that special. But w3schools earned their reputation when that was exceptionally rude. Also, straight up wrong all over the place. But that is more of a symptom. And to be fair a lot of the software back then was straight up wrong, too. Tough job documenting the reality of the WWW correctly while IE was around.
Which I prefer; I got burned by Neeva shutting down shortly after they started (for a pretty mild version of the word "burned", it cost me very little, but it was annoying).
The only tech merch I've ever worn was back when firefox had a good logo, goodwill, and no text on the front of the shirt.
Lots of people just want a simple graphic tee and IFKYK. It's not sports. Bands can sometimes get exceptions, but often approach wrestling tshirt levels of gaudiness.
Tangentally, IIRC the setting for the next Elder Scrolls game was all but confirmed after some creative at Bethesda accidentally made public a Pinterest collection with weapons and armor from the medieval arab world.
Some lies are actually better than the truth. My favorite example is that shooters are taught to imagine their sights being in line with their gun and that's they're shooting horizontally. But they're not, they're shooting in an arc so over short distances bullet just rises a little. Don't ask why. Why it's cool is that people using this model are better shots.
I am an average internet user.
The average Internet user doesn't visit HN at all.
The intersection of these groups, HN visitors that comment? Distinctly not average.
;-)
https://en.wikipedia.org/wiki/Many_a_true_word_is_spoken_in_...
Even as an Arch user, probably at least 80% of my usage of the Arch Wiki is just going through "Troubleshooting" list of previously seen issues and solutions for whatever thing I'm dealing with. I don't go in expecting that everything will work exactly the same for me, but over the years I've ended up with quite a few headaches solved by pasting the right line in the right conf from one of those sections.
Maybe the value comes not just because they bother to maintain a wiki, but that Arch Linux tends to select for above-average technical people, even more than Linux in general does. (Even if Arch people strangely don't run Debian goodness; but we benefit from a diversity of perspectives, even if they are unexplained. :)
I am the admin of two FB groups, both UK home education related (one about exams and qualifications, the other for single parents). I do it because I feel I should help the community (especially as the crappy commercial groups that target the same audience), which is made up of non-technically inclined, mostly middle aged women - i.e. FB's core demographic. I cannot remember the other demographic nos offhand, but its 95% women in the exams group, and 98% in the other (and both have 1% other/did not say), and the age profile reflects the fact that people have school age (or just over, in the 16 to 18 age group) children.
I had actually planned to try and push the community towards forums, but with the Online Safety Act in force that is not a risk I am inclined to take.
My cousin runs a family group where she shares photos and wishes people happy birthday. I would prefer her to use WhatsApp but she is the one running it.
A lot of my friends post family news, important things like births and deaths and weddings on FB.
I understand what you're saying, but my own observation is "many of us" == "a tiny minority". "There because our friends our communities are" is exactly it...our friends and communities don't care as long as FB and the other social media cesspools are the default. I too have tried to get various relations to not make me log into social media to be part of their life, and I get some variation of "oh, you weird little infosec guy, that's way too much work and it's good enough for all of us".
I don’t use it, but the Privacy Pass [0] thing should actually make it great for privacy if you care more about that than personalization.
Looking at their privacy policy they state the following:
> We may store web requests made by user browser temporarily, with strict retention periods, for debugging purposes, and in a manner that they are not linked to an account.
For 2 years they've said "Kagi has zero marketing budget and spends zero on marketing", that's a lie and they still haven't changed it. They've sponsored over 200 youtube videos https://filmot.com/search/%22brought%20to%20you%20in%20assoc...
I asked because the search you've linked to only seems to list relatively niche chess videos on one creator's account. And the videos themselves are also ad-supported. It's (at least) not impossible that the sponsorship association is not budgeted in the way that (I think) you infer.
I think the bigger danger is the competition around AI chat assistants. There are other paid assistants, but you already see that even the paid assistants are getting trained to promote certain corporations.
All the ones I liked were pinterest posts with zero attribution. A reverse image search then just brings up dozens of ripped and reposted copies of that pinterest post, also without attribution.
It gets frustrating
https://kagifeedback.org/d/3822-kagi-is-sponsoring-youtube-v...
"I watched Daniel King for 10 years and have previously donated to him, now I am happy that we are in position to sponsor his work which I believe is extraordinary.", this also definitely sounds like monetary compensation to me.
It was incredibly disheartening to have no recourse to attribute my own work, other than to smear some gross watermark on it. The automatic removal of that metadata, along with AI image generation, are some of the reasons why I gave up on the hobby entirely.
It's incredibly hard and stressful to derive any sort of pleasure or interest from something when the second it's exposed to the internet, any sense of humanity you tried to attach to it is stripped away, burned, and commercialized for the monetary benefit of some ethereal financier. It's the sound of an invisible vacuum cleaner, whisking away any sense of joy or life you wanted to share with the world for common love; the death of sharing. For-pay hugs.