Per user it’s about 15 queries per day. I’m sure there will be some that are incredibly active and some that aren’t active at all, but 15 per day seems quite reasonable.
I keep trying chatgpt style products, but rarely reach for them because I don’t really have a concrete use case for them. I use ? all the time, code completion at work and regularly generate images. Generic chat can do those things but is never as good as specialized tools I have access to.
My guess is that purpose-built specialized expert AIs are going to end up being more useful than the “everything” products like the chatgpt UI, but time will tell, I guess.
I have no numbers but I bet my # of searches/day went down drastically because of that alone.
What the billing page doesn't go into detail about is number of searches caused by The Assistant, number of FastGPT searches, and number of regular searches. I'm curious because I'd like to track my tendencies; whether I am slowly using AI more than search, or if it stays the same. And FastGPT is a grayzone.
I genuinely thought it would be higher, but I suppose bang patterns don't count.
(Posting this mostly so that people who are curious about subscribing to Kagi can get a sense of how many queries they're likely to need to use.)
Date (UTC) AI Tokens AI Cost (USD) Searches
Jun 2025 0 0.000 141
May 2025 0 0.000 743
Apr 2025 0 0.000 723
Mar 2025 0 0.000 621
Feb 2025 0 0.000 556
Jan 2025 10,692 0.000 1,189
Dec 2024 0 0.000 805> We are also thrilled to report that we have achieved profitability. This significant milestone is a testament to our sustainable growth and fiscal responsibility. It demonstrates that our approach of offering a premium, ad-free search experience resonates with users who support a service aligning with their values. Becoming profitable allows us to reinvest in the business, further enhancing our offerings and ensuring that we can continue to provide a top-notch search experience.
As long as they're profitable I don't mind at all if they stay small. They're extremely useful to me as it is, and their small size means they aren't targeted for SEO nonsense, so their methods to cut through all that still actually work in my experience.
Not every business needs to become a unicorn. Some businesses are better at small scales serving a specific niche, and by their report Kagi seems to have found their niche.
Obviously they don't have the ancillary services (Maps etc), but for just searching, Kagi is far more likely to surface useful results instead of just the highest bidder. Compare a search like "us esta" for a clear demonstration.
First search result for "us esta" in Google is "https://esta.cbp.dhs.gov/esta", same as in kagi, is that your objectivite fail at coming up with a simple metric?
Does ublock or ddg etc block sponsored results?
I also have my own WTFs with Google, I even started collecting them: https://imgur.com/a/bgFax59
First one on Kagi is the official US government ESTA website.
Let others handle rarer queries.
I'm not entirely sure what "objectively superior" is even supposed to mean in the context of a search engine, or how this follows from having 50k users, but that ceases to be an even remotely plausible statement if you've ever attempted to get good non-English search results.
On the index side I agree, I think they are using other people indexes so far, I don't know if they are thinking about building one themselves.
Same for LLM, but I think that there the problem is even worse.
Here's a list for the curious:
>A look at search engines with their own indexes
https://seirdy.one/posts/2021/03/10/search-engines-with-own-...
https://help.kagi.com/kagi/search-details/search-sources.htm...
It also uses other indices along Google.
> All results from external indexes.
The above is something I see all the time when using Kagi.
Unsurprisingly it's Pintrest that annoys people the most.
I haven't used Kagi much, but I don't think I've ever seen a single Pinterest result from any other search engine, I barely even know what the thing is for.
Is this a US thing? A Kagi quirk specifically?
And to be honest, I don't see as much Pintrest crap as I used to. It may just be that it has fallen out of favour because of its terrible design. But it was a site that basically let you create mood boards from images that you found as your browsed.
The big problem with it was that it simply copied the image so if you were looking for anything, Pintrest came up with the images for a product but then had removed all the source so it just flooded search results with uncited references and you couldn't find anything useful.
Reverse image searching to find sites thst sold a prpduct became useless.
If search works well, the result you wanted is in the first spot on page #1. This is a bad outcome for Google.
The more follow up searches you run and the more trash you wade through, the more ads you see and the more money Google makes.
Then you have to pick a finger in the air multiplier for the value of the business. A stockmarket listed SaaS company that isn't over-inflated might be 10x the revenue, so that would be $50 million valuation.
Kagi is small, but it must still have good margins. So maybe really it is 5x revenue in value, depends on lots of things! Who selling to, and predictions for long term growth.
Edit: And the insane values these days are P/E of 90+, bad businesses are less than 5, so I took a conservative estimate of 10 P/E, but I think a more reasonable number might actually be 40, putting them in the centimillion category for sure.
Kagi probably have a user base of users who are highly attached to the product’s quality. Kagi could lose most of their paying customers should they ever fall into the wrong hands.
But I’m glad it’s like this. A good old company that just sell good products to their happy customers.
[1]: https://kagi.peopleforce.io/careers/v/125936-core-back-end-t...
I doubt you can get a feeling for the work / life balance from this half sentence.
> Our ambition is enormous, going against industry giants with a very small team. You will work a lot. We are completely user funded. Kagi is currently used by one town worth of people. Do not expect VC backed/big-tech salary. Do expect equity as a part of compensation
In other words, maybe they’re saying they don’t have any BS jobs like Meta and Google seem to have.
Google’s HR review team that reviews the team that reviews interview processes is probably larger than Kagi.
Most startups just go through the cycle of cheap and great - hit the profitability button and turn into a flaming pile of crap.
It used to be such a good service. Beautiful docs. An interface that made sense. Great support. Now it’s the very definition of flaming pile of crap.
Before it became the example of how to invoke hatred in a software team.
If you retrain yourself to work the way Jira does, and use all the defaults, it’s not bad at all at what it does. Quite good.
But if you use it as a bug tracker only, or customize it to all the business processes you’ve evolved over twenty years, it becomes a frightening morass.
He also put all hardware and software issues into the same sprint "to work as one team", except the hardware issues had very little to do with the software issues; also, the hardware people never updated their tickets, so each sprint just had the same 40-50 spam messages for which you had to create custom filters to avoid.
He also changed the issue sizing mechanism once in a while. So we'd have hours, t-shirts, and Fibonacci numbers (including some odd non-Fibonacci numbers that "seemed right").
I would always prefer a less feature-rich issue tracker with sane defaults.
Linear.app, GitHub Projects, post-its on the back wall of an antisocial project lead, anything other than Jira. It just attracts people who think "Category" is a good category.
Kago can do this because they bootstrapped it.
From their website:
Kagi was bootstrapped from 2018 to 2023 with ~$3M initial funding from the founder. In 2023, Kagi raised $670K from Kagi users in its first external fundraise, followed by $1.88M raised in 2024, again from our users, bringing the number of users-investors to 93.
Kagi launched in June 2022 and we maintain a public page tracking real-time Kagi growth and usage statistics at kagi.com/stats.
In early 2024, Kagi became a Public Benefit Corporation (PBC).
> Every program attempts to expand until it can read mail. Those programs which cannot so expand are replaced by ones which can.
Or the newer version:
> My point was not about copycats, it was about platformization. Apps that you "live in" all day have pressure to become everything and do everything. An app for editing text becomes an IDE, then an OS. An app for displaying hypertext documents becomes a mail reader, then an OS.
So in turn, every product becomes bloatware that needs more money to maintain and more users to get more money.
he said, typing from Firefox with its new "open AI chatbot" sidebar button
Businesses rarely remain stable, no matter if they're startups or not. Because that wouldn't make any sense. Either they shrink or they grow. You can call this the law of midrange businesses.
Consider a midrange hotel:
Either the owner cares about his business and continually improves the facilities and the experience for the guests. Soon the hotel will have a good reputation and will constantly be full. So the natural next step is for him to increase prices, because there is the demand and also he has higher operating costs. Repeat this process over the years and the midrange hotel is a high-end hotel.
Or the owner does not care about his business and continually lets things decay and become a worse experience for the guests. Maybe because he wants to save on operating and investment costs. Soon the hotel will have a bad reputation and the owner will decrease prices to attract guests, then further cut costs because cheaper guests don't demand much. Repeat this process over the years and the midrange hotel is now a low-end hotel.
And this happens in all businesses, because in the end they are run by people. If you'd love for companies to get profitable and remain small-ish, then you have to make such a company.
Honestly, I find this whole startup mentality, where you only build a company so that you might later sell it off to some megacorp, very strange and off-putting. It essentially means you didn't care about your product and your users in the first place.
If you can bootstrap it yourself then there's no need to do this, but those that bring in investors will need an exit.
You can do two, or even all three of those things. Human beings are not boolean greed machines.
The HN bubble likes to reduce everything to a numbers game. Real life isn't like that, as demonstrated by the many tens of thousands of companies that aren't run like a dystopian Silicon Valley comic book.
You just never hear about them because they’re small and like the businesses they invest in, they’re satisfied with moderate returns.
From their site:
Kagi was bootstrapped from 2018 to 2023 with ~$3M initial funding from the founder. In 2023, Kagi raised $670K from Kagi users in its first external fundraise, followed by $1.88M raised in 2024, again from our users, bringing the number of users-investors to 93.
Kagi launched in June 2022 and we maintain a public page tracking real-time Kagi growth and usage statistics at kagi.com/stats.
In early 2024, Kagi became a Public Benefit Corporation (PBC).
Key difference: They didn’t take any funding from anyone whose values are misaligned with their own
These are legal and contractual proceedings, with strict definitions of terms by law, not by popular opinion in the hacker community.
"Underlying motivations" of the investors is never a legal factor in any kind of investment deal. That's not something that can be accounted for by any kind of contract.
I would not have guessed that. Why would I need a VC firm to invest a few thousand dollars into a company?
They raised money in the form of SAFE notes, in which case a venture capital firm is created (you can also call it a legal entity or a juridical person) and each investor owns a part of that firm in proportion to the money they invest. That firm in turn will have a contract with the startup company which details and regulates how the invested funds in the future can be transformed to actual shares.
American companies, startups in particular, have terrible support. It's really nice to have actual contact with someone when an issue arises.
However, is there really no other model than these two, i.e. being 100% the product vs. paying the "full price" of the search?
I guess the problem probably has to do with the value of the GPU cycles being lower than the served content. This is most apparent in the case of AI; e.g. if the mining lasts as long as the session, and the server runs 2 GPUs while the user is only running 1, then you can’t complete the “payment” unless the mining continues beyond the length of the session.
Though training data sets based on private information about people are worth more to them, so they focus on that.
That lets them do things like run crooked ad auctions that screw websites and advertisers, intentionally worsen search results exactly enough to maximize profits, and other stuff that came out during their trial.
That problem has been solved for anybody who wants it solved, through Apple Pay and Google Pay, or even the built-in feature in the browser for remembering credit cards.
Normal people absolutely would not accept websites hi-jacking their computer to mine crypto and hardware manufacturers like Apple would swiftly implement measures to protect users from those freaks.
Do people really use web search that little? According to the stats Kagi shows me, I make about 50-100 web searches every day on average.
And no, there is only those 2 solutions.
I'm still mad that they took away the '+' operator for that turd of a project. [0]
[0] To be clear, it totally could have been a great project. Early on, there were signs that it was going to be -at worst- decent. But, well, Vivek Gundotra wanted the project to be a big turd, so it ended up being a big turd.
Somehow, Google decided to show me the ferry timetable: https://imgur.com/a/bgFax59
Like, whut?
It seems that Kagi isn't expecting that things will change much [2].
[1] https://www.wired.com/story/bing-microsoft-api-support-endin...
[2] https://kagifeedback.org/d/7107-microsoft-bing-retiring-sear...
I get that they want to replace Google, but removing such a bang is user-hostile and I'm unhappy about it for those 2 times a month I need to compare with Google's results.
EDIT: nvm it works iff you add the bang at the start of the query, for some reason.
I always put it at the end (you do a normal query, and add the bang the second time around) which used to work on DDG which introduced them.
5usd in america aren't 5 usd in the rest of the world indeed
My lattes are far less expensive than that.
I know I'm speaking form a position of privilege and this will be unpopular; but I'm not fond of subsidizing other users by paying premium prices for my subscriptions.
As someone in the EU/Schengen, I need results for 3 languages or 4 regions (if "worldwide" counts as one region) on a weekly basis. If a Dutch payment method would mean getting only Dutch results, it would be about as valuable as a newspaper with all pages about trade and international affairs blacked out: you'd need to buy it for every region you're interested in which probably surpasses the original price
TBH, if i where them i'd be trying to serve open source models from my own infra, much cheaper to pay per GPU's per hour and batch process all your users prompts, than leave that big 95% fat margins to OpenAI and Anthropic
But I guess they have customers who want those APIs anyways, idk, again, i thought they where a search service, not an ai company, so this sub for llms business deal is weird from that POV? like great that it works for them to get money/customers but that doesnt seem their main point of existing?
Kagi’s official position is not to support regional pricing (visit https://kagifeedback.org/ and search for “regional pricing” to go through the “Implement regional pricing” thread). The service is probably out of reach even for many people in the first world. Even its family tier is expensive.
Hopefully, when it reaches a much higher number of users, it’s able to reduce prices. Or it can just remain a niche service and potentially be disrupted by a competitor.
Skipping past the top two sponsored results in DDG really isn't that big a deal and still diversifies the search market from Google's monopoly, so paying the equivalent of a streaming subscription service... idk. None of my friends seem interested in paying for what is currently free (on top of the hassle of being logged in everywhere all the time) so sharing a duo/family account isn't an option for me either. Maybe that's something that would put it in reach for you?
Edit: apparently they've shared what it costs them to provide the service:
> a single search costs us 1.5 cents to deliver
https://kagifeedback.org/d/687-implement-regional-pricing/23
They charge 54€$/year for 3600 searches, which is.... precisely 1.5 cents per search. That sounds a little bit too convenient that the claimed cost price is precisely what they charge
This is either not the cost price or relies on people not using their subscription, otherwise you could never recoup R&D costs. Maybe this is total expenses divided by number of searches being run? I.e., it includes wages for the devs working on new features, their administration, etc. (fixed costs) and isn't actually the expense that each additional search incurs to deliver
DDG displays up to 2 ads above search results (most of the time I get none: trying just now, I saw 1 ad when doing two product searches and one knowledge search). The most favorable figure for 1000 ad impressions is 6$ according to <https://spideraf.com/learning-hub/what-is-the-average-cost-p...>. That's 0.6 cents per ad if the advertising network costs nothing. Advertising on DDG goes via Microsoft, so they'll take a cut: <https://duckduckgo.com/duckduckgo-help-pages/company/adverti...>. Guessing that they take 10% (I expect it's more), you're looking at 0.2 cents average revenue for each search I did just now. (Not a reliable figure but as a ballpark estimate.) DDG must have much lower costs somehow (maybe they just take results from Bing verbatim and have few costs of their own)
I switched from DDG. It saves me more than $10/month, both in time and in actual dollar costs due to suboptimal search engine results. It’s as simple as that.
The jump in improvement from google to ddg is probably bigger than the jump from ddg to kagi though. Also, I switched before ddg and kagi had AI search. Kagi’s AI is a huge differentiator for me, and I haven’t used the DDG one that much.
I also can't associate searches I do for work with a personal account, and installing special software (this privacy pass client) is also iffy, so that's not going to work for people like me
Otherwise I'll have to use a publicly accessible search engine all the time anyway, since there's no way I can get my boss to pay for a search engine when I can't even get him to pay for our primary communications mechanism (calls and chat) that is hosted by a third party relying on donations
And that €54/$54 is the price when you pay per year, if you pay per month you’re paying more per search (although at least part of that extra money will go towards handling the payments)
If nothing else, Google taught me that just because something is great today doesn't mean it will necessarily be great tomorrow. I can't get attached.
Soon those users will stop caring about the product that they use, but rather about the value of the underlying share/s.
There exists exactly 0 business structures that can solve the problem of 100.00% of a userbase being completely satisfied with all decision-making.
However they have been on that course for a good part of the last year though, and they ultimately deliver a good product, which is what matters at the end of the day. Kagi mostly feels "feature-complete", and I'd rather have them spend time on the AI projects than trying to be too inventive and overloading the core product, which is the route many other startups take when they get to that point.
So I think it’s fair for them to at least have people thinking about that. Plus, the features they have aren’t intrusive and are completely optional. Like, it’d be dumb for a company to not spend development time on a threat that has a decent potential to shrink their (already captured) market.
Sometimes I search for things I know I am looking for. Other times I don’t know quite what I’m looking for or I know in advance that I’m not likely to find it—so I chuck it at Llama 4 Maverick and it usually gives me something useful.
I had no plans to use the LLMs until they opened it up on my tier. At this point however, it’s half the value I get out of Kagi.
Also am worried about the move to mail, I already have fastmail, and kagi would need to create a heck of a mail client for me to even consider switching. I'd much rather have a company that does search very well, a company that does mail very well, and a good communication between the two.
And I also have less tangible worries about Vlad's demeanor when I see some of his writing in the feedback forum or discord. It comes across as ambitious but not very circumspect, but maybe that's what's necessary to make it in this sector. I won't pretend to have enough experience to offer much opinion on the matter, all I can say is that its unsettling at times.
Out of curiosity, what communication do you want to see between your mail service and search service? Half of what I'd like to get out of paying is keeping them separate!
I'll rephrase to state that I'd much rather have a unix-like philosophy where you have small-ish companies each specializing and being very good at one thing and then you have an active medium they all live in and talk together and hold hands and...
Hopefully, now that they've proven that you can create a profitable company from doing this, more will follow and we'll get a proper ecosystem.
Like others, I'd prefer to see Kagi succeed at being the best possible search engine and nothing else, than try and do the Google thing and do everything.
I really like their browser, Orion though. It's still rough, and crashes at times but it feels like a great non-chromium option for MacOS.
Features like boosting the niche forums I browse for search results is just a bonus on top.
I agree that I could go back to DDG and not feel like I am missing too much, but that doesn't bother me.
I agree there is not a lot of differentiation between stock Kagi search and DDG. DDG still has a few ads but it's not that annoying, perfectly usable.
Kagi's assistants are pretty interesting though. Recently I asked it to find back a post that I vaguely remembered. It managed to generate a bunch of Kagi searches with different keywords and narrow it down to an old tweet.
It does indeed, as was revealed during the Google Search antitrust case in the DDG testimony.
The extension is a hack that allows them to redirect the searches from one of the available search engines to Kagi. So first you search, then e.g. the DDG page loads halfway, then the page flashes, you get redirected to Kagi. If you are lucky, you are signed in, and you wait for the Kagi results. If not, you sign in and type your search again.
On iOS, I made a shortcut that pops up a native text prompt for your search, then opens kagi using your session token. I recommend this over the extension.
On desktop Safari, you can additionally make Kagi your home page. Not as convenient as searching directly from the top bar but not bad either.
The extension is a hack that allows them to redirect the searches from one of the available search engines to Kagi. So first you search, then e.g. the DDG page loads halfway, then the page flashes, you get redirected to Kagi. If you are lucky, you are signed in, and you wait for the Kagi results. If not, you sign in and type your search again.
On iOS, I made a shortcut that pops up a native text prompt for your search, then opens Kagi using your session token. I recommend this over the extension, it is fast snd reliable.
I replaced it with a "let's fetch" shortcut.
Wikipedia first. No Pinterest, no W3Schools, no Fox etc etc.
They have a nice list for suggestions.
The results aren't necessarily better than other search engines in general but the personalization is so incredibly valuable.
Oh and having it auto rewrite Reddit results to Old Reddit helps a lot too.
Plus, it seems like having to login is inevitably makes it easier for them to associate activity with a specific real-like person. It's probably easier for them to associate my activity with me when I log in than it is for DDG or Google to track my activity when I'm not logged in and can't be easily distinguished from the dozen other people that used that computer that day.
Yes, this is annoying. They make it as easy as they can (QR code login, session links) but it's still a speedbump.
> having to login is inevitably makes it easier for them to associate activity with a specific real-like person.
That's a legitimate concern. Kagi added Privacy Pass support to mitigate it
I stopped using duckduckgo after they censored tankman: https://news.ycombinator.com/item?id=27394925
I am tired of being advertising meat, and I'm willing to pay to use the services that do not waste my time.
A real pity.
2.) Switching to "All Data" also doesn't set the y-axis to zero, but to 6,840.
Having an y-axis set not to zero is in the majority a sign of people who want to inflate growth.
The reason to have a non-zero y-axis for time series is to amplify changes, e.g. the changes might be to small to see with a zeroed y-axis. Or you have ups and downs and want to compare them, with a zeroed y-axis again the changes might be too small to compare.
Whenever you want to show growth, a non-zero y-axis is usually a sign that the aim is to overstate growth, because we as humans estimate growth by the steepness of the graph, not by the numbers. A non-zero y-axis creates a much steeper graph and thus growth is perceived much higher than it is.
Not zero indexing is misleading if you are comparing discrete things like GPU performance, not in the case of plotting a timeline graph. Their published stats could be seen as misleading if they only displayed a short and/or a specific timeline (excluding the latest data for example).
I'm tired about seeing all those posts about Kagi. I have yet to see a single example where it outperforms Google. People just don't know how to search.
Every time someone claims Google is becoming bad, ask them to share what they're looking for and how they search for it, and where the problem lies becomes glaringly obvious.
My original point is that Kagi is NOT better than Google at search.
Also simply bad faith.
Check my response to this article and the thread of comments that ensued for a pretty recent example: https://news.ycombinator.com/item?id=43829655
I have never ever had a problem with search. If I can't find something on Google it means it simply doesn't exist.
With my ad-blocker enabled, I was able to replicate your results for the "expedited passport renewal" search. (state.gov right under the AI Overview)
Then I disabled my ad-blocker, and the sponsored results and AI Overview pushed the first actual result (still state.gov) below the fold.
Actually for me the AI overview does show up but BELOW the official government's website (and there's no ad at all).
To be fair I think Google has lowered their position a lot in the results because it used to be that it would be 80% Pinterest results and now even in images it's only a few here and there.
I think they understood that everyone hates that website.
Migrate fully, after Google first search turned into a cesspool of scam AD's and SEO crap.
They go so far as to have an API to generate cryptographic proof of subscription tokens without revealing your identity for searching when using Tor, etc ( https://help.kagi.com/kagi/privacy/tor.html ).
Beyond their data collection stance, which I believe, their results are better and worth paying for because they don't have all the extra ads and crap shoved in them and allow you to modify your own website rankings, etc.
As long as I cannot access their data, cannot see their systems on a continuous basis, they can collect my data. The same applies to DuckDuckGo and the same applies to VPN services.
So at Kagi they are aware of Tor, they are aware of the Dark Web. If you even talk about this on your website, of course you find ways to circumvent privacy there.
I tested Kagi and my experience is the results are even with or worse than what other services present like Google or Bing or even Perplexity. And if you don't like advertising, just use an ad blocker. I'm so surprised that so many people don't seem to do this and they always complain about ads. Just use an ad blocker already.
So they allow me to modify my own website ranking. Well, if almost nobody uses this website, that's pointless. Besides, if I could modify my website ranking, so can everybody else. And now we are back to search engine optimization.
I don't understand all of your objections, but it sounds like you're operating in a very strict "if I can't verify it myself continually, then I don't trust it at all", which I think is far too black-and-white. Trust exists on a scale, is built up slowly over time, depends on your threat model, etc.
I don't want an adversarial relationship with my search engine where they try to show me enough ads to make money to stay afloat and I try to block enough ads so my experience isn't miserable - this isn't any fun, and one of us will lose.
On the other hand, I can pay Kagi, and they are incentivized to give me the best service possible so that I continue paying them and recommend them to others. It's far more positive, and a better outcome for us both. And it's working - I have a much better experience on Kagi than Google or elsewhere, and I have kept paying them and recommending them to others entirely organically - I don't think there's even a referral program, I'm certainly not using one.
They say they don't collect my data, and why would they? They don't show ads, and if anyone ever found out they were lying their paying customers would leave in droves. Furthermore, they take actions to back up their claims, like integrating the Privacy Pass feature.
Finally, they don't allow you to modify your website ranking for others - they let you modify how other websites rank for you. You get to personalize how you want websites to be ranked for you when you make a search, which is a very cool feature.
Anyway, if you tried it and didn't like the search results, no problem! You can use something else. I don't think that your other objections are valid, though.
Local deep searches with ollama are comparable in quality and cost nothing.
Also, half of their premium model list is laughable - 32/70b models can be run locally without an issue.
I found that have two main use cases:
1) Search - It is nice that Kagi search is less SEO spammed. That a big appeal. However, I don't reach for a search engine when I have a topical question anymore. I reach for something like o3 with search, so I don't need to dig through a bunch of articles. Over the last few years my pure search volume has dramatically reduced.
2) Maps - I travel a lot. When I toss a business into the search bar, 90% of the time I want to see it on a map. Google's Maps experience is just far superior to Kagi's. I find that I end up typing maps.google.com more often than I am happy Kagi ran the search.
I don't mind paying for a product; it just didn't work for me.
Side note: I personally dislike that their favicon looks like a "g" for google. It's always confused me.
I also find significantly less "vibe written" smut on Kagi. I'm not sure if they are targeting it specifically, or if it's getting corralled by some other metric (like ads+trackers).
For my personal workflow, I don’t see the appeal of having a separate setup if I’m just going to use the same underlying models anyway.
2) Yeah this is the one thing I still use google for. Though they did recently update Kagi maps (a weeka ago I think?).
On AI - I think they have no choice tbh, they need to bring something to the table there. I'm pretty pleased with their AI implementation. In search it only activates if you append a question mark. Their assistant is a pretty good alternative chat interface, it lets you choose the model. I cancelled my chatgpt sub because I can use kagi for many models. It has probably fallen behind the tooling others have at this point though.
> You like the service but don't agree with the company spending money on... t-shirts [...] It smells a bit like you want to control what they spend on?
i like the service, but spending money on t-shirts seems unreasonable to me. I'm sure there are a dozen things which would benefit from the money and time which were spent on manufacturing clothes. From their own blog post [1]:
> The process from here involves setting up a business entity in Germany, so we can import the t-shirts, store them in a warehouse, connect inventory logistics and ship them all over the world. This includes building a website and connecting it to a back-end database
which sounds to me like not the best way to spend the company's time
also:
> why did we go through all this trouble and allocate nearly a third of our investor-raised funds to produce and freely distribute 20,000 t-shirts
all their answers to that question they asked themselves in the same blog post seem silly to me. Go give bonuses to your employees, upgrade devices, make a company event, etc
> On AI - I think they have no choice tbh, they need to bring something to the table there
absolutely not! If i want an AI search result, i go to an AI provider of choice. Again, i don't think that should be their focus
> I cancelled my chatgpt sub because I can use kagi for many models
replace chatgpt with claude and reverse the statement and it will be true for me — i cancelled Kagi subscription because i use claude as a search engine
> replace chatgpt with claude and reverse the statement and it will be true for me
... like, ok. But now I'm confused. You didn't want them doing AI but you're not opposed to it. With kagi I can use chatgpt, claude, others. But you don't need to pay for the level that includes that so once again I'm puzzled as to what your point is.
Kagi is not for you - fine, I'd have no problem with that. But the way you've written your comments is that you do like the service but you didn't like what the company did in other ways... not a moral objection but that it wasn't, in your view, an efficient use of their time. And I'm absolutely sure this isn't what you meant which is why I've asked questions about it.
you are dead right, but spending nearly a third of the total revenue sounds like a waste for me. They aren’t an established company, they are still very small in terms of users. I would be way happier seeing them re-investing in growth. If they would signal somewhere at any point they want to keep being small, i would totally get their move with the t-shirts. Since they haven’t mentioned it even once to my best knowledge, i assume they are taking the traditional path which is growth
> You didn't want them doing AI but you're not opposed to it. [...] But you don't need to pay for the level that includes that so once again I'm puzzled as to what your point is
i pay for the pro version which has CLI support, it’s crucial for me and lets me to almost never leave my natural habitat, terminal. Also it has bigger context windows, bigger token limits, better everything, which i don’t believe i can achieve with Kagi’s subscription. I’m strong proponent of the Unix philosophy: each thing should do only one thing and do it well. To me AI, just like t-shirts, is an unaffordable luxury which a company like Kagi at the moment absolutely shouldn’t even look at. You may argue that in the modern world AI seems like a next logical evolutional step of search engines — but to me is trying to be a jack of all trades, master of few. I paid for the product for 2 years, i expected different things. Give me bigger limits of include/exclude website lists in lenses, maintain your own blacklist of garbage websites like pinterest (you can even use AI for this), create a public collection of easily discoverable lenses, give better local searches like the ones google provide. I didn’t get any of that, and then i received an email about my t-shirt being ready to be shipped — and shipping wasn’t free by the way, which was confusing to me personally after so much noise about free t-shirts. That subtle detail, still paid shipping, was another putting off factor, which seemed like a dark marketing pattern
> I'm absolutely sure this isn't what you meant
sorry, i didn’t get that, can you please explain what you are sure about that i didn’t mean something?
But from my first days experience I can safely say that 300 searches a month is a low number for entry pricing. And given that I'm in a developing country, the entry pricing is also not cheap enough.
But I'll keep pushing for a couple of months.
moralestapia•8mo ago
People truly don't gaf about anything as long as they're getting something for free. Sad.
I assume this is paying users, if this is total MAU ... damn.
PS. I do not mean Kagi is a bad product. I think they're great, actually. I'm just complaining about how poor the reception has been.
TylerE•8mo ago
dymk•8mo ago
sshine•8mo ago
Which means I can’t embed Kagi Search or Assistant into my open source products because APIs are second-class citizens.
To be fair, it seems like this makes sense for them economically, and maybe it even makes sense for growing in the right direction.
But as someone who would rather use their API, it’s a pity Search and Assistant aren’t there.
112233•8mo ago
Once I tried openrouter, I am not touching any subscription LLM providers.
Google/Microsoft/Apple "all-in-one" personal subscriptions are a different beast, because they charge well below cost.
baobabKoodaa•8mo ago
moralestapia•8mo ago
sshine•8mo ago
sshine•8mo ago
And what matters to me is: that it’s enough to keep going, securing the service, and that the number is steadily increasing.
dymk•8mo ago
sshine•8mo ago
moralestapia•8mo ago
How do you know?
My napkin math says they're in red numbers, but let's see yours! :)
sjs382•8mo ago
Profitable as of a year ago
sshine•8mo ago
They reached profitability one year ago, two years into their existence. It’s a cool company.
Edit: Corrected from two years to one year
moralestapia•8mo ago
~Buddy, we are still in 2025 ...~ (removed, parent comment corrected)
If they were profitable then why did they raise money again?
>inb4, to expand their business
If that's the reason, where is this expansion? Their growth curve looks exactly the same since they started ...
IMO they invested a lot of money into their AI efforts, so maybe they raised for R&D?
I want them to do well, it's just that I thought they already had a much larger market share.
sshine•8mo ago
They hired a technical architect and an email engineer, so the biggest expansion is clearly into email hosting. But I imagine that AI R&D continues. I hope they consider more agentic extensions than search.
UberFly•8mo ago
bruce511•8mo ago
As much as we like to complain about ads, we also aren't really prepared to fund things that aren't ad driven.
Frankly on the web, for me, ads are fine. For TV less so. I pay for streaming (no ads) and for sport (no ad interruptions) etc.
For search, half the time I'm clicking on the ad anyway.
xvilka•8mo ago
omnimus•8mo ago
jan_g•8mo ago
sshine•8mo ago
Aurornis•8mo ago
I think Kagi is great. I also understand why most people don’t need Kagi. It’s not hard to see why most people aren’t interested in yet another monthly payment on top of all the other things we’re asked to pay for right now.
sshine•8mo ago
drawfloat•8mo ago
sshine•8mo ago
Well, unless you're working for government and next year's budget depends on this year's expenses.
Most modern people need search engines. And most people don't care about advertising and having their private data sold. And most of those who do are happy to use adblockers and/or free alternatives to Google that still show ads, but may or may not collect as much data about you.
So it is indeed a niche: People who want to pay for search and AI, when both of those are "free".
As for having an AI broker instead of getting it straight from the model companies: I think the economic incentives are better. I don't want bloated answers, just because the model companies make money per token. I want to either pay for each token, or pay a lump sum to somebody who optimizes the result for me and pockets the difference.
hotmeals•8mo ago
eviks•8mo ago