Hacker News new | past | comments | ask | show | jobs | submit login
Kagi: A Premium Search Engine (kagi.com)
373 points by bcg361 on Jan 7, 2022 | hide | past | favorite | 221 comments



Founder here. Thanks for submitting it, but I am afraid there is not much to see right now.

We are still in closed beta (mostly to control cost and gradually improve user experience), with much more left to do. There are some examples of what Kagi can do in my twitter feed. [1]

Main idea of Kagi is to offer a paid search product, where user is also the customer and is central to everything that happens. Btw. Kagi is pronounced ka-gi (similar to yogi), I get asked that a lot so there it is...

If you want to beta test, feel free to sign up and we will do our best to send an invite within a week.

[1] https://twitter.com/vladquant


Couple of things:

First, price point is 2-4x what I expected. Not saying I wouldn’t be a customer but at $5 unlimited it would be a no brainer for me.

Related: I would love to see family pricing. $12-13 price point is just right for me to share with my family with up to six people or some such.

Privacy is important. Is it possible to provide a payment system that is anonymous? Cash in the mail? BTC or similar? What is your policy on logging searches?

Will your results ever feature ads? Affiliate links? Will you ever attempt to create Kagi Plus social network?


Unfortunately we are not in the position to set the price point by consensus or market expectation but by realistic cost of providing the service at a given level of quality, in a way that potentially ensures sustainability and serving the customers long-term.


>we are not in the position to set the price point by... market expectation

While I do like the project and wish you the best, producing a product that the market will not pay for is a business model doomed to fail.


You will find a lot more people willing to pay with a price drop. $10/month is not possible for folks outside the first world.


I agree - I like the idea and so far have liked the beta. But $10 a month is way too much unless there is something that really makes it THAT much better than other search engines. Look at NextDNS for example - they offer it for free under a certain amount of queries and for home use unlimited is 2$ a month. And it has a lot of features, easy and clean interface, plus custom options. I gladly pay for that because its cheap and I can use it with devices off my home network.

$10 a month is getting into the territory of streaming video services and those require a hell of a lot of bandwidth and infrastructure to maintain.

For business or corporate use though $10 a month / user seems totally applicable.


I disagree. I would paid $10 / month as an individual.


+1 for anonymous payments (Monero please?). I am signed up for the beta with a non-personal email but using traditional payments will throw that semi-anonymity out of the window.

What subscription models exist in the crypto world right now, i.e. to automatically charge me every month?


Echoing the family plan, I would love to have my family use the search engine but at $10+ per person, it doesn't make sense.


Is there a way to prove that i had paid yet remain anonymous (ideally without a session attached to me)? You offer pay-per-use or pay-per-month or both?

Also feedback regarding your current website:

So there is merely one sentence: It says

"Premium search engine where everything on the page matters."

Yet, to the left and the right of this sentence are wiggling clouds that don't matter at all and distract the reader. Not ideal.


After you log in however it is closer to true.

Personally I'd almost say I can live fine with not pleasing extreme nitpickers but I think they'll fix it.

The important point for me is unlike Google and DDG almost every result on Kagi does contain the thing you search for and if not you can report a bug and get a reply from a real human who will update the behavior. I've seen it multiple times after I joined their chat.


There is a lot more on the current website than that sentence and the clouds.


OP is contesting the claim that everything on the page matters. If anything on the page doesn't matter, the claim is wrong.


Your application of logic is not valid here, because "everything on the page matters" has not been given in a formal way, and you're cherry-picking one interpretation.

What if it is intended to mean something like:

  ∀x ∃u: onpage(x) →  matters(x, u)
"For each element x, there is some user u, such that if x is on the page, it matters to u."

This is not contradicted by this version of "anything doesn't matter":

  ∃z ∃v: onpage(z) ∧ ¬ matters(z, v)
"There exist elements z for which there are users v, such that z is on the page, yet doesn't matter to v."

But is contradicted by a different version of "anything doesn't matter" like:

  ∃z ∀v : onpage(x) ∧ ¬ matters(z, v)
But you can easily find a person in the world who doesn't care about any aspect of something, let alone just one aspect of it. So what?

If we restrict x to just the universe of users who are interested in at least one thing on the page (i.e. who are actual users), and we have somehow ascertained that everything on the page is covered by at least one user in that set, then the claim holds. You need to supply actual evidence that there is an item which no user ("person interested in at least one thing") is interested in; just you not being interested in something isn't it.


>>If anything on the page doesn't matter, the claim is wrong.

Define 'matters' and 'matters to who'.

May not matter to you, but if research shows that people interested in the reading about the product stay's around longer if the page is visually appealing - then it matters to someone.


Knock knock.

Who's there?

To.

To who?

To whom!


It all matters to me. And if it matters to anyone it matters. So the OP must be right.


The landing page is technically not a search engine, but maybe “every result matters” is better phrasing here.


Hi Vladimir,

Thanks for the invite.

Please start monetizing : we want to support this effort ( I understand it is a difficult decision ). Would subscribe if reasonable.

Register a new HN handle here for comms about Kagi.

I see gimples of the future where my search engine behaves/looks differently depending if I am researching IT issues or looking for clothing.

Good luck.


> Register a new HN handle here for comms about Kagi.

I think one of the best things about HN is that there are few "corporate" accounts, and founders continue to communicate from their personal accounts.


I agree with you, though most of those handles do in some way relate to the person and are typically not 'fantasy style'.

I didn't mean to suggest 'register hi-kagi-ceo-here'. Maybe 'The Vlad' would work.


vlad_the_imputer

Sort of works. Apologies for the joke, it’s the Friday of the first week back.


Cool product. If I was in your innovation team you know what idea I would suggest? Adversarial compatibility!

Support indexing internal stuff for enterprises. Let's say I have a kagi enterprise account and my company had integrations setup. When I search for something, I would get results from internal wiki, email , sharepoint, etc... along with internet results. One stop shop for search.

I hope you read this and give it serious consideration. You wanna kill google? This is how!


> You wanna kill google? This is how!

We don't want to kill Google :) Google serves a purpose in the world and it did help enable the modern society to exist, with all its marvels and flaws. Heck, it even enables Kagi to exist!

Think of Kagi as a small premium brand, providing a very different search experience.

Enterprise search is something that we are remotely thinking about, but frankly we want to focus on providing the best possible consumer experience first, so that individuals and families around the world can enjoy a user-centric, ad-free search experience.


Thanks for the reply. I am glad you are focused on the mission at hand. I hope you will prove a profitable ad-free search engine is feasible . If you're not going to kill google, at least I hope you humble them by having better quality.


This is a great idea and lets face it - enterprise subscriptions are where the real money is at.


Hi Freediver,

Would you mind telling us a little of the story about Kagi the company behind the product?

It’s really interesting to hear that you’re using Crystal. Where are you guys based and how have you bootstraped?

Good luck and keep at it, this is a really interesting project.


Sure.

We are running on a tight budget, completely bootstrapped by funds from my previous exit. Trying to control cost and all that while establishing product-market fit, which is the reason the beta is closed for now.

Instead of starting with technology, I started by selecting people that I can work with and are crazy enough to accept working on such an ambitious idea. It turned out that they really liked Crystal and, so, here we are. I have over 30 years of dev experience, and in my previous startup I wrote a significant part of the code. But never wrote a line of code in Crystal before so it does make me feel a little uncomfortable. Its performance is truly amazing and that is what I care most about. Our front-end payload is HTML + vanilla JS (which btw is not needed, everything works in no-js), and probably about 100k on average, very lightweight.

I am based in SF Bay Area, and we are a fully remote team.

Thanks for your interest!


As someone who happily pays for many "free" products/services (like DNS service through NextDNS), I am super excited about products like yours, where I can get a better product and more/better features for money. I received my invite this morning and I am super excited to check it out. I wish you all the best of luck and I am really rooting for your success!


+1, happy paying NextDNS customer here (for 1.5 years now) and recent Kagi user, very happy so far (likely future customer)!


Love NextDNS as well - my only gripe with it is that I cant have separate filters and lists for different sets of devices. Would love to have my family on some more basic security and privacy stuff while keeping the advanced and higher false positive based stuff for my devices.


Hmm, what do you mean? I have a number of different profiles, all with unique settings (lists). It's to the right of the NextDNS logo in the UI as a dropdown. Then in the client you choose which profile you want to connect to.


I guess for the most part I am not using any the clients - I just point my home router DNS, my Android operating system DNS, and any browsers that might bypass. I erroneously assumed those profiles could only be chosen if the client was being used but it looks like I may actually be able to create different profiles and then still just point the devices network settings to the correct IP. Too bad for "LinkedIP" they dont let you manually specify (it looks like you have to click the button while you are on a machine using the network you want to link)


Hmm yeah I was actually also just pointing IP's for a good while[0] until my mobile ISP (4G router) decided to not allow custom DNS to be used, after which I installed the NextDNS app (with DoH or whatever is default) and got rid of that restriction.

[0] I remember I had a cronjob calling a particular custom NextDNS URL regularly, through which I'd like to remember I was able to choose a specific profile


I've not seen much production use of crystal, but its cool to see such a nice product using it. I've been quite happy with kagi since I got my invite.

Crystal really is quite neat, I've got a few years experience with go and picking up Crystal has been easy and quick to get productive. Being able to call C so easily (especially when generated with crystal_lib) is pretty amazing and a big boost compared with go.


great to see Crystal getting adopted in more production grade projects, honestly, It would be awesome to see the language receiving more traction in 2022


So how do I apply?


How would you solve the issue of self-censorship when the exact identity of a searcher is known?

Can you Guarantee the privacy of searches and that you will never sell search history data?


Those are good questions, thank you.

I think that a class of people, who ask kind of questions that need absolute guarantee of anonymity (I am thinking terrorists, drug dealers, traffickers etc) are:

- Looking for a different thing (because anonymity is not the same as privacy)

- Are probably looking at tools that cost much more than Kagi does

We are not even interested in serving that market.

Having said that:

- Kagi does not store search queries nor by definition it is then possible to associate something that does not exist with an account

- In general, Kagi does not store or mine user data in any way. Our privacy policy is pretty detailed, I recommend checking it out https://kagi.com/privacy

The strongest guarantee I can realistically give you that the above is true, is that our business model is simply not dependent on mining or selling user data. And the premise of our entire business would fall apart if the opposite turned out to be true. In other words the alignment of incentives here is as good as it gets.

We are not interested in your identity or searches, we are interested in providing the best search product for you, so we can keep you as a paying customer as long as we can. The moment either of the above is not true, you will walk away. It is wonderful to be in this position from a product/business perspective.


I would expect 99% of us not wanting our search history revealed or just even logged. Specially with English as a second language, I've searched for things (to learn what they are) that are just horrible, think blue pancake/muffin/whatever it was.

Or just plainly searching about legal or medical issues to learn/get informed can yield some very questionable queries taken out of context as well.

Do you currently log user search queries?


I understand the pedagogical example as a stark contrast, but this choice is immediately turning me off the product. Startups and the companies they grow into have spent decades helping themselves to data to track and identify and write clauses into their policies and terms of use to be able to stockpile behavioral data for a potential future "pivot" if nothing else.

We are living in a world where it is more based in fact to assume that companies will help themselves to this kind of information under a variety of pretenses than that they won't. The only reason people will assume differently is if they trust the company, the product and the people behind it. The policy seems to be doing its part, but trust needs to be built by every conversation with a potential future customer. If the question that pops up is "how can I be anonymous if I pay you each month", the resulting argumentation should not be able to be construed as "we're sorry, but we're not going to help be your burner phone", which the comment I'm replying to is toeing dangerously close to.

People search for many things and it could detail their interests, their current location, their financial troubles. The commitment needs to be "we will never, never, never, ever, do anything like this", and not just "it doesn't make sense for us to do this". Because if someone wanted to build in that sort of tracking, it could "make sense" from a business standpoint to do this in that the data, if collected, has a lot of value on the market.

Even if the incentives are indeed aligned to keep the paying customer pleased, we also need to know that if we do walk away, that at that point there will not possibly be anything left as an artifact that a future buyer could do anything with. This assurance will be most effective if it's rooted in trust and values rather than in practical concerns; the practical concerns are valid, but only if they are restraints that have been applied in search of an objective by yourself, instead of "it is not currently in our business plan". (As an extreme example, consider a bank saying "we'd never pick items out of our customers' safe deposit boxes and sell them; it would simply be too much work".)

Basically, the facts seem reasonable enough. But you need to work on not coming off like the parent saying "well I'm sorry you want to hide things from me", which is what unprompted bringing up the covering up of criminal activity in response to questions of anonymity and privacy does. I understand that those concerns do come up in thinking responsibly about a product like this from all angles, but making it a part of a discussion with a potential customer disrespects and denigrates their fundamental needs enormously, the same needs that would make them attracted to the product in the first place.


I understand I could have better communicated my message and thanks for bringing it up.

My point is that none of "us" require anonymity in normal circumstances of using a search engine. But we all absolutely require our privacy respected (and many people still conflate privacy with anonymity).

Kagi is 100% privacy respecting by default, and examples I gave illustrate not only that, but that certain (almost total) level of anonymity can be obtained by Kagi too (because searches are not logged, so they can not be logged with an identity). In a paid service environment where you need to authenticate the user, help them restore their account etc, it is very difficult, I dare say impossible, to achieve total anonymity guarantee from a technical standpoint and I want to be transparent about it.

I think a better sort of guarantee is exactly the one that I gave - our business model does not incentivize and sort of privacy invasive behavior.

> The commitment needs to be "we will never, never, never, ever, do anything like this"

Perhaps you have not read our privacy policy, but this is exactly what is says there, and I took it for granted. This is probably my mistake and I should have assumed that people submitting inquiry will not have read it.

In my mind "there is no need for us to do it" is a much more powerful driving force than simply stating we won't. Stating something is easy but not substantial as witnessed by many big tech companies. The reality is that their business model forces them to twist reality away from privacy statements like those. Our business model does exactly the opposite, and has a positive feedback loop reinforcement in relation with the customer (we cheat you -> you take your wallet elsewhere).

Rereading this very answer I have a feeling it will not be the most satisfying again, so I hope that with your help, we can come to an answer that will satisfy your inquiry.


> How would you solve the issue of self-censorship when the exact identity of a searcher is known?

What do you mean with that?

> Can you Guarantee the privacy of searches and that you will never sell search history data?

I think that’s literally impossible to do?

FWIW, selling data would probably not be worth it, and ruin them. It’s supposed to be pretty expensive (early discussions are around $10-$20 a month) while only having a tiny amount of users (10s of thousands). At least for me, that aligns their incentives properly. And of course, they are bootstrapped without any VC money.


Self-censorship is the behavior where one does not perform certain searches because their interest in the topic will become known to others. This might include health information, sexual interests, political views or other sensitive topics. The fact that the search engine has your name, address, and payment info means that they can associate you with 100% certainty to everything you search for which can have negative consequences for the user under different scenarios.

In regard to selling the data, I've seen businesses where they sell user info just as a side project to balance the sheets. It will b an issue only if one is caught. Somewhere there I looked around the website but I couldn't even find where is the HQ and under which jurisdiction they operate. The privacy policy had lots of technical details, but not legal one.

It would be interesting if like Signal they can validate an user but provenly cannot associate the account with any actions. This might be an issue for billing and tiered plans though.


I pinged their devs, especially regarding their location. I’d assume US as that’s where their founder is according to his personal website [0], but I agree it would be better to have that information there.

Thanks for the explanation. I can only repeat that I don’t think this is solvable. Unlike Signal, which uses E2E this is not possible here, so whatever promises Kagi makes will always have to stay promises as they’ll need to know the content of your search and for what account it is (as your settings influence the results) to give the right reply. At least that’s how I understand it.

[0]: https://vladimir.prelovac.com/


I think it it solvable:

1st Allow users to share the token by which they are identified. 2nd Implement a 3rd party website that offers a 'free tier' via token roulette.

That will generate noise in search results, making it hard to pin a particular search on someone. Optimally users would be able to decide how much 'noise generating searches' they allow.

Off course that would only work if paying users would be volume / rate limited.


Could potentially go the Mulvad approach and allow mailing in of cash for an anonymous user ID (numbers only) and then you enforce DoT and DoH everywhere? Kagi itself would still have access to your search queries and could identify you buy your public IP but it would be a decent step in the right direction


Thanks, it would be interesting to see new solutions in the space.


I'm also interested in this, especially since they will have the user's payment information with the search history.


Hi Vladamir.

If it is possible can you share what is your tech stack specially your database ?


How did you manage to crawl the web without any issues?

Did you get any angry letters and/or lawyery threats to stop crawling a specific site?


Since you seem to be using it – what are your thoughts on the Crystal language? Pleasant? Ready for production?


I know that the team is having lots of fun with Crystal and the performance is truly amazing. It allows us to run the entire setup very economically.


Congratulations for what you accomplished so far.

Question: I have a little "vanity website" of my oen. Is there a way to make sure it gets included (in full or maybe just some pages) in Kagi searches?


Check Teclis [1], this is a demo front to one of our noncommercial indexes, and if your site is such, you can submit it there.

[1] http://teclis.com


Consider https://webmonetization.org/


k-ah-gi as in tawdry

k-ay-gi as in lady

k-aa-gi as in laggy

Which one is it?


It is ka-gi, like yogi.


So... ko-gi? The phoneme used by the letter o in yogi is not something anyone is going to associate with the letter a.

There are multiple phonemes often represented by the letter a; the GP is asking which of those is correct.

Edit: all of these assume American English pronunciations.


The explanation that kogi is pronounced like yogi implied to me, limited being that I am, that the 'g' was a so-called hard consonant, not that the vowel sounded like 'oh'.

But yes, that does assume American English pronunciation (and for what I know, various other Englishes too).


No sorry for confusion. It is ka-gi, just accented in the same way as yogi.


"Kah-Gi" or "Kar-Gi"?


Perhaps this settles it: https://www.youtube.com/watch?v=ig4VTr0rt4Q


Thanks!


This is awesome. If it works well, I will pay for it without a moments hesitation.


How is it better than using a proxy server to access Google?


or whoogle: https://github.com/benbusby/whoogle-search (which i love btw)


If I can be frank, nothing google has done or is doing with my search data bothers me in the slightest. If it uses my data to help target advertising - so be it. It doesn't bother me either as I use uBlock. So whether I pay or not, the only thing I care about is the quality of my search results. What makes you think you can do better than a company with over 20 years experience, 10s of 1000s of engineers working on their product? I took a skim through the features and google could implement site muting in an instant if they wanted - so this isn't going to get you far in the market. The number one thing that makes google better than anyone else at this is I type in some words and they find the information I need - pretty quickly.

I usually wouldn't ask these kinds of questions but I think this is the one of the hardest problems to get right - to impact such a wide number of people. No-one has even come _close_ to taking a dent out of Google in the last 20 years.


> What makes you think you can do better than a company with over 20 years experience, 10s of 1000s of engineers working on their product?

I've tested it since December.

What made them think they could do it I don't know but they already outclass Google by a wide margin.

The shortest and most precise explanation I can give is:

Kagi is to mainstream search engines today what Google was to mainstream search engines 20 years ago.

And that maybe settles your question too?

> No-one has even come _close_ to taking a dent out of Google in the last 20 years.

Unluckily for power users (and maybe ordinary users) and luckily for would-be Google competitors Google has rested on their laurels, told everyone the problem was webspam while their quality declined down the the levels of earlier search engines. Edit: Kagi shows the problem is not webspam but Googles insistence on 1.) not letting people block domains 2.) not respecting peoples queries.

Competing with Google a decade ago would be hard.

DDG still isn't anywhere near peak Google quality and yet they've seen (actual) exponential growth for years. That wouldn't have happened if Google still was leagues ahead.

Until December I started in DDG and used !g to see if Google had a different result that was better. Often it did not.

Since December I have used Google twice to verify something from Kagi or to look up something Kagi couldn't find (I'll try to get a bug report underway freediver, I just need to recreate using non personal data.)


>DDG still isn't anywhere near peak Google quality and yet they've seen (actual) exponential growth for years.

Google gets the total of DDG's lifetime searches (over a decade's worth) in about half a week. "Exponential growth" is overselling it.


Maybe they’re using “exponential” metaphorically to mean “a lot”. But if they’re using “exponential” literally, then maybe the exponent is 1.03.


Look at the graphs.

To me it looks literally exponential until half a year ago.


or look here: https://news.ycombinator.com/item?id=29852783

Seems I'm not alone in thinking it has been almost exponential for a good while and until recently.


Other people may have different opinions about their personal data.

The advantage Kagi has over Google is the different business model. Google would never for example introduce a "non-commercial" filter as it would go against their advertising-based business model. Kagi doesn't give a shit about ads and will do whatever that keeps its users paying and so has a non-commercial "lens" (as they call it). I'm not sure how good or accurate it is, but it's at least there which is already more than I can say about Google.

Kagi has a feature to prioritize or exclude some domains. Google doesn't have that despite their size, resources and decades of search experience. That makes it pretty clear to me that Google's weakness has nothing to do with experience or resources and thus a new entrant could potentially compete without either.


If Google went completely add free, they would need to charge $57 per year to entire US population. I bet vast majority of population will rather be willing to endure ads. HN bubble often forgets how small it is relative to whole population.


The price point of Kagi is anywhere from $120/yr to $240/yr, based on their current FAQ.

Let's say they manage to take just 1% of Google's daily search traffic, which based on a quick Google search (hah) would be about 35 MILLION search queries. If we do a very very rough ballpark estimate of 50 queries per user, per day, it would be about 700,000 unique users... or between 84,000,000 and 168,000,000 USD/year gross revenue.

Now, do I think it's reasonable to expect them to take 1% of Google's search traffic anytime soon? Probably not. I do however think that the sheer size of this market means there is a LOT of missed opportunity by the incumbents in the market that can be leveraged to create niche products that perform successfully by any other market's standards.

So, even if the HN bubble is small there are other bubbles out there. A LOT of other bubbles out there.


>which based on a quick Google search (hah) would be about 35 MILLION search queries

Public numbers are a decade old. You're off by an order of magnitude.


That's fair enough, but the option should be available for those who are willing to pay. $57 per year sounds like a steal for anyone doing any kind of productivity work with computers.


I think it is worse. The people who are willing to pay $57 to opt out of ads are probably the most profitable (I am guessing here: no citation available). Seems like a case of Adverse selection https://en.wikipedia.org/wiki/Adverse_selection


That doesn't seem obvious to me. I'd be more inclined to think anyone willing to pay to avoid ads likely isn't very receptive to advertising in the first place. If someone is so annoyed by ads, they're probably not going to be clicking on them, right?


In terms of purchasing power they might look desirable, but surely they’re also likely to be massively overrepresented as users of ad blocking solutions (also pure speculation on my part). That could make them less profitable in practice.


In my case, Google doesn't work too well.

I usually have something in mind when I search, and Google doesn't offer any customization for me to put it in. Instead, it shows me what it prefers.

For example, I frequently need to switch languages and/or regions. You can't do that with Google (except maybe logging in and switching some settings but lmao, I need a switch). No matter what I do, Google reads my IP range and "decides" what results I should view. Very frequently, this means I literally can not find the thing I am looking for.

For many people - I assume anyone who isn't American - Google simply isn't working very well. We use it, yes. But DuckDuckGo has a Region switch that already makes it 10x more useful than Google and usually offers far superior results. A new search engine could expand on that, offering more customizations.

I'd go so far that for every human, except 350 million Americans who never switch language or region, Google is ALREADY the inferior search engine and people use it because it is what they know.


The infuriating thing is that Google used to do this right. It used to be possible to go to Google.de and see the German results for instance. Now I always get the Swedish results (I live in Sweden). I often have searches where I want results from Sweden, Germany Australia/NZ or general English. I have to become increasingly elaborate with my search terms to get them (essentially search with whole sentences in the desired language), it's really annoying and I'm surprised how Google could get the so wrong.


I appreciate the sentiment and I can certainly understand it. This is of course what I thought too in the beginning, but as curiosity and then determination led one thing to another, we learned that things are doable, as much as that sounded as science fiction. Certainly there is evidence in user feedback that, at least for some people, Kagi is decent at what it does.

The fundamental premise that enables it is the focus on the user. It allows truly wonderful things to happen in the product (and we sometimes forget that search is still a product). Aligning the incentives between us and the user (which is also the customer) is really freeing from a product management perspective.

We are still users of Google products - we utilize the paid search API for results and GCP for infrastructure (hello Google!). We think there are some things that can be built with these truly amazing assets, that Google's business model simply does not permit, and we are thankful for the opportunity.

In the meantime we also built our own index which focuses only on noncommercial content and are going to invest more in this direction in the future. We call it "the more humane web".

Another thing is that we do not need to get _very_ far in the market. There is no shareholder pressure, just angry (or happy!) users. This is not, nor will ever be, a Google scale product.

This is a paid search product of premium quality (or so I'd like to believe it is shaping to be) for a tight number of individuals (and families!) who are a little different and want something different for their search experience. Few tens of thousands of subscribers will make Kagi sustainable and I hope we will have what it takes to get there. That is all I can wish for right now.

Thanks for considering!


Towards that end, I wonder if you could partner with or incorporate some of the things Marginalia is doing. https://search.marginalia.nu/


I do plan on offer some sort of API access, but I just haven't had the time to work out the details yet. In general I do believe in collaboration within the alt-search space, but I'm a bit reluctant to have my non-profit work end up being the back-end for someone else's for-profit operation. Still thinking about it.

In part I've been postponing this because I'm planning a bigger overhaul of rate limiting and access stuff, in part to get rid off cloudflare because I don't want a creepy MITM spying on my users if it isn't absolutely necessary.


Thanks for the reply. Let's see what happens and I wish you all the best!


> If it uses my data to help target advertising - so be it

What a great thing that data leaks and power abuse don't ever happen. /s

> What makes you think you can do better than a company with over 20 years experience

The severe decline in quality? I've been using DDG for a while now; in about 2 years I've looked at Google results maybe 5 times and was disappointed every time. If you think Google is the best you can get right now you're missing out. Your argument is based on the assumption that a company with more experience cannot possibly be bested at what they're doing.

> google could implement site muting in an instant if they wanted

Google could do a lot of great things, but unused potential is worthless. Remember how Google got big precisely because they did what competitors could do but didn't?


Google has been getting worse for the past few years, probably in the demands for making a standard product to suit the average person. But now the first page is full of results that exclude one of the search terms. And my strategy for years to find obscure results- of finding the oddest set of relevant terms to describe what I want to find to weed out commonplace results that share a couple terms — has stopped working almost entirely as Google's algo seems to have upweighted popularity considerably.

At this point I don't think it would be that hard to provide a useful alternative, especially with a business model where the customer is the searcher and not the advertiser. I'm not sure that all of those engineers are having an additive affect on the UI. Quite the contrary, I imagine many are working toward improving the experience for advertisers, who are the customers.


The quality of cadbury chocolate has been getting worse over the years since Kraft bought them out. I still buy it because it looks like the same product, kinda tastes like it, and ultimately will give me what I want. I don't disagree the quality of Google's results has gone down - but you are talking about changing user behaviour here and for the majority (not you and I) google still does what it needs.


It may even be improving for most people's needs. But I think it does present a scenario where a niche search engine with a different business model could be successful (and they probably wouldn't need 10k engineers to reach parity with Google's, at least in that niche). Maybe there isn't enough of mes though.

I haven't changed search engines, and Bing won hands down when I did the Bing/Google challenge they promoted in 2012. But it is getting 'bad enough' and I need search results for projects enough where I'd try the beta if invited and would probably stick with it if it's a good experience.

Hell, if they gave me a few more Scopus-like tools for more precise queries on a version of Google scholar, I'd be sold.


The problem I have is that Google is not optimizing for the best results - they are optimizing for homogeneous "best" results for the average person. And a lot of that comes down to optimizing for showing product placements (even if they are not ads) and difficulty filtering out garbage paperclip sites and repetitive blogs and whatnot. It used to be that I could fine tune my google searches using just quotes or other simple "advanced" methods and get great results but that has significantly decreased as the years have gone by. If I search for a string in quotes I dont want google automatically "suggesting" I meant something else and then I have to click another button that I am "sure" I meant my original terms.


This argument always comes up. What makes you think you can... Bla bla bla.

When Google search came, they started their search engine by providing a clean search page with no ads, at a time when all the others had cluttered, ad-filled, shitty pages.

Users picked Google because it was the better experience back then. Today, if a search engine comes along that gives you great search results and privacy, it will become very popular quickly, because a lot of us are fed up with Google and want them to go away completely.


Google came along when the search industry was in its infancy, with multiple providers vying from top-spot. We have now had over 20 years of dominance, refinements to get it to what it is today. Entering the market today is on a whole different spectrum to what it was back then.


If anyone is able to take a large bite out of Google's search market share, it will be by figuring out how to not drown results in ads, SEO and blogspam. If they can successfully pull that off, it has obvious value.


Easy to detect ads and affiliate links, just down rank all commercial sites. The good sites are all free anyway.


Quality of google results, has been going down for quite a while.

There have been some speculation that it's down to SEO and google trying to make as much money as they can.

It's possible that a company which does not derive their income from search ADS could create a better search engine.

If someone can make a better search, I am all for it, even if its just for limited topics.


Subjectively google's results have got dramatically worse over the past 10 years. Whether this is intentional on google's part because they have decided to prioritise ad revenue over useful results, or if it is because the problem has got much harder because of the increased number of pages and the increased sophistication of spammers is open to question.

If the answer is a) then Kagi or another disruptor could undoubtedly produce better search results, and probably in a short period of time, with a small number of decent engineers as there is now a vast body of research and praxis to build on that didn't exist when google was starting up. If the answer is some variant on b) then they're on a hiding to nothing.


You don’t care what Google does (aside from search) because you consciously installed a tool which hides their work from you. Glorious self-deception!

By hiding ads, you’re forcing everyone else to pay for your search. I’m glad I use DDG, and I?3 signed up for Kagi’s beta.


Been using them for about a week, very happy with it.

Regional results (especially since the last update) feel as automatic as google. While I still hope they’ll add my suggestion of a bang for region switches, most queries I do don’t need one, and it properly switches between English and German.

All of the DDG bangs. Seriosuly, search without bangs is nothing I’d ever want to go back to.

Built-In Domain Blacklist. Goodbye Pinterest. Forever.

General results are great. More relevant results than either DDG or Google and a great presentation (like showing snippets from accepted or upvoted SO answers)

And I have not even played around with advanced filters like lenses (which people in their discord seem to love).

Negatives: So far, I’d say only that the widgets are less usable than DDGs.

I had low expectations, but I’m really very positively surprised.


I've been in the beta for a while and I find Kagi results to be way ahead of anyone else. I've made it the default search in Alfred so that I have instant Kagi in Safari.


I've been using Kagi for a couple months and generally like it. It's the default search engine in my browser.

My only complaint is that occasionally I run a search and get zero results, like the page is literally blank (I try the search again and it is still blank). I then try the search on Google and get what I'm looking for. I guess this happens because Kagi's index is smaller, but I don't really know.


It happens because there is a bug that we are still trying to find :) Rest assured we will find it! (it is probably somewhere in the way we feed feed data through socket for maximum performance and something either on GCP end or our client/server code is 'too' optimized)


Agreed, I'm also very happy so far and only had to use !g twice so far.

Interestingly enough I haven't actually seen that much cross-language results (also DE/EN), even for ambiguous queries.


What's a "bang"?


!imdb for a direct IMDB search, !rt for Rotten Tomatoes, !g for google.

DDG calls [0] them bangs, and as I have been using them in recent years, I use that term everywhere now.

[0]: https://duckduckgo.com/bang


It's the "!" character that you put in front of certain things, like "!g" ("bang gee") for searching Google through DuckDuckGo.


The second character of the shebang

    #!


Also known as hashbang.


Whoa... this was really confusing to me for a moment. Kagi.com used to be one of the premier digital commerce platforms for software and shareware, especially Mac games, serving customers from 1994 until 2016. [0][1][2] Anyone who registered one of Ambrosia's awesome shareware games used Kagi's services, for example. It's really weird to see the company name and domain return again in this way! Best of luck to the team to live up to this legendary name :)

[0] https://web.archive.org/web/19961216050944/http://www.kagi.c...

[1] https://web.archive.org/web/20160413151203/https://www.kagi....

[2] https://tidbits.com/2016/08/04/kagi-shuts-down-after-falling...


This brought back memories. The first money I ever made from software as a teenager was through Kagi. It was some Mac utility shareware apps written in REALbasic around the year 2000.


Nice, that's awesome! How did you market/publish your software? Just a website? I remember trying to make a Hotline client/bot with REALbasic. Part of me definitely wishes I had pursued that stuff more strongly, maybe I could have authored some nice shareware as well!


REALbasic? Hmm, if it was about 2006 I’d wonder if it was some of my creations.

Long time since I did shareware in any language, though. Something like £2k total takings, I wasn’t good at business.


(And now I realise I misread the comment as spent money rather than made money)


#nopressure


The name reminded me of Kägi, a Swiss chocolate wafer: https://www.kaegi.com/en/


Yes! This was my first though - I didn't realise they had gone under. I think I bought at least one app via them.


Yeah, that was my thought too


There have been a couple of discussions on HN lately [0][1] about the current state of search engine results, be it served by Google, Bing, DDG or other. Also, it seems like there is a lot of work on search engine alternatives as in the case of Kagi and also here [2] to name but a few. Of all alternatives, Kagi seems to be off to a very good start and I personally am a very happy beta user.

What I have not seen so far being discussed anywhere, though, is the Willingness to Pay [3] for a search engine alternative, say like Kagi. What do you think is an acceptable monthly subscription price? What do you think is an expensive price? What do you think is a prohibitively expensive price?

Here is my take on those questions: 5$, 10$, 20$.

UPDATE: I'll summarize all answers and post the results here for reference.

[0] https://news.ycombinator.com/item?id=29782186

[1] https://news.ycombinator.com/item?id=29772136

[2] https://news.ycombinator.com/item?id=29690877

[3] https://en.wikipedia.org/wiki/Willingness_to_pay


Not enough answers to be representative, but here is what we have so far:

     % Customers who find the price acceptable
       ^
       |
  100% |------|
       |      |------|
   80% |             |
       |             |------|
   60% |                    |
       |                    |
   40% |                    |
       |                    |-------------|
   20% |                                  |
       |                                  |
       -------|------|------|------|------|-->

       0$     5$     8$    10$    12$    20$

                  Price per month
UPDATE: Upvoting another person's answer is something that I cannot differentiate, so I cannot include it in the summary above.


After testing it for a month I say $10 is a steal for me. $20 I'll probably pay. Above that it depends very much.

Although I hate dumb subscriptions for stuff that should be paid once and for all I have a history of not only saying I would pay for certain thing but also doing it.

And with Kagi, if they keep respecting my queries and have a sane price: it is a no brainer.


Honestly, I'd say around $2-5. It's not that I wouldn't value better quality search but in a world where search feels like an almost 'invisible' part of browsing the web I'd struggle to pay that much. I guess it's comparable to what I feel is a price I'd be willing to pay for a VPN say compared to a coffee subscription.

This also isn't a reflection of what I percieve the work in building a search engine to be but more my subconcious perception of what I'm getting.


For purely search, 10$ is about the max I would pay.

For a full equivalent to Google's services including Maps, Docs, Drive, etc, I'd be happy to go much higher up to 50$.


Superhuman is $30.

So upper end might be higher for an "apple like" slice of the user base.


There’s a huge difference between what I’d be willing to pay for a good search engine and how much I feel should be charged amid competition (and cost to provide). Existing search engines are good enough for most purposes, but if there was a good enough alternative I’d be willing to pay a pretty substantial markup (and if google continues getting worse or the search engine gets better…)

10-20$ a month for a meaningfully better search engine is pennies compared to the possible value it could provide (if it provides search results of sufficient quality). Every search I have to spend a couple minutes refining saved would make it worth it.


Interestingly, people are apparently willing to pay $1500 per month to forgo using search engines [0], implying this is what the service is worth to them. I guess here you're asking for price on privacy, as Google is already free to use.

edit: I guess even the range of answers here, compared to implied value of the service, is an indication how much people (me included) value their data.

[0] https://www.economist.com/graphic-detail/2018/04/25/how-much...


I think the wording in your post is wrong - That economist article says people would want to be paid $1,500 if they had to sacrifice access to a search engine (not that people would pay to sacrifice access).


I would say for general purposes (casual everyday searches) - anything between 5$ and 10$ / month would be a good price for me. This is of course dependent on the quality of the results - it has to be a real service and not a marketing platform. Most of the single user licenses for web services like github, dropbox are between 5-10$, so the search engine had better deliver quality results, if it requires payments.


I think $120/yr is good. I pay $120/yr for Dropbox, which is also a service I use multiple times a day and I think it's fair.


Same for me, although I thought about Spotify who get 10 Euro/month from me.


I already pay over 100$ a month for subscriptions I rarely use. If we get alternative to Google that is simply on par or slightly better, 10$ is nobrainer just for the de-googlification. For something significantly better - like option of paying some extra cash for more processing power - 20$.


I could imagine paying even up to $100+ but only for niche use-cases (finance data, fiding experts/employees). I would like ability to create plugins for search usecases and for example take 20% cut, basically allow for search app store. I wonder if it would work.


The idea of search plugins for specific use-cases sounds incredibly incredibly interesting!


Around 100 usd on annual subscription basis.


It seems the trend lately for new services to be built with rather high monthly fees.

From the FAQ: "We plan to offer entry level plans for as low as $10/month, unlimited plan at around $20/mo as well as make bundles (to include Kagi email and other services), family plans and annual payment discounts available."

I have to say I support a lot of these initiatives, but I can't help but think altogether they're creating a division of the web into premium services the rich can afford, and low quality services for the masses. Most of the world doesn't have the silicon valley salaries where $10/month is considered low. I feel like there should be an ideological consideration as well to make better services accessible to as many people as possible. Just my 2 cents.


Not really sure that you need a 'silicon valley salary' to afford $10/month.

Netflix charges almost double this and they have 100's of millions of subscribers worldwide; my guess is that a premium search engine only needs a small fraction of that number of users to become a worthwhile and profitable service.

Can't say if I would pay for it or not - haven't used it yet - but if that was the price of dumping google searches I for one would be OK with it.


Actually the comparison with Netflix is quite apt. For twice the price I get lots (and I really mean lots) of shows streamed to my house. In contrast here, here I get a search service that is provided by others for free. For the average Joe the value proposition is very different. So you really have to be valuing privacy very strongly and/or have significant disposible income.

Note I actually like the service and I signed up for the beta, but I understand the GPs point about building two "webs" one for the masses and one for the "well off" that can afford these tools. However as one other poster said, you have to bootstrap a service somehow and if it gets popular enough there could be the ability to reduce fees to give it mass appeal.


There is I think one distinction in the analogy to be made. We shall not put the sign of equation between services offered by Kagi and Google, because we are not for example doing the same for Netflix and Youtube.

One may pay Netflix because there are _specific_ shows that they can not get anywhere else, or simply because the quality and convenience is much higher than if they tried to watch those or other shows in different ways, including free.

In similar fashion, Kagi offers a completely different product/search experience to Google, with different results, different features and attention to detail like you would expect from a premium service obsessed with user experience. Saying that Google is offering same thing like Kagi but for free, is ignoring this nuance and is same as saying that all streamed video content is the same, or that every comedy show is the same.


I see value in paying for a service which would curate results by eliminating content farms.

Average Joe would also not see any value as he doesn't know what a content farm is to begin with.

It's all about finding a niche (and selling out once you're on track to get traction, but alas).


And how many people share their logins to avoid having a bunch of fees from different services? I bet its more common on the lower end of the income spectrum.

When you're making 15/hour with a family things are pretty tight


> When you're making 15/hour with a family things are pretty tight

Correct. But what Volvo invented (3 point safety belt and more) soon became standard in ordinary cars.

Say Kagi is a luxury product for some people: it will still open the road for others to see that it is possible to carve out a niche that isn't as crowded as ad-sponsored-idiotic-AI-manipulated-answers-for-everyone-and-noone-at-the-same-time.

Neeva is freemium.

I think you.com is freemium too.

Marginalia is free and at this point so good that it is sometimes a better fallback from DDG than Google is. (To be fair, Marginalia is a side project running on a single desktop pc in Sweden somewhere but Google is that bad these days that if DDG doesn't have you covered chanses are Google doesn't either. And then marginalia comes in with a delightfully different ranking, respecting my queries and allowing me to find things - if they exists in its index.)


> but I can't help but think altogether they're creating a division of the web into premium services the rich can afford

If I may, there is huge complexity at hand here. It is one thing to offer a meditation app for $2/mo, where its peak complexity is changing colors on the screen. It is entire different to be able to offer the entire internet searchable in less than 500ms in the palm of your hand. The ability to offer that at $20-30/mo with robust privacy protection is frankly a miracle.

Second factor, is that Kagi is bootstrapped and our priorities are quality of services offered, not the number of users. And knowing what I know, it is simply not economically viable to do this at a lower price point and hope for sustainability of the business.

Finally, the price is only one side of equation, the other is value. The question is will you be getting enough value for your money? For me, the ability for my children to use a search engine without being bombarded with ads from an early age has a lot of value. I already pay $15/mo for YouTube Premium for the same reason. I understand that not every parent is in this position to choose, but many are.

Having said all that I do not pretend that Kagi is designed to be for everyone, like Google is. It is a premium service for a reason - the cost of providing the quality that we provide is simply what it is, and we clearly said no to the ads from the get-go. Remember that the main reason most of the mainstream web is free are ads.

The ideological consideration you refer to is something I agree with, but that role is on governments to fullfill IMO. I do not think they will anytime soon so I am doing what I can. Access to information like access to water should be part of public infrastructure and we miss the latter in the modern society (as public libraries are not enough anymore). We can easily get into the realm of politics here so I choose to stop :)


Can you talk about why the price needs to be at the 20-30$ level and what infrastructure and engineering needs to be built / maintained / expanded on to run a search engine?


Here is a simplified example. Bing API is $7/1000 queries. Most users with HN pedigree search 100+ times a day. A single search for an error message can produce 10 follow up searches. That is $21/mo+ just in API cost a month. And we have lot of users hitting 200 queries/day limit (imposed in beta), so they would be hitting $40+/mo in API cost alone when we go live. This is cost without factoring infrastructure costs, other operating costs and salaries.

In order to have a hope of becoming sustainable you need to make a calculation that with a certain number of users you can break even.


If more people would pay it would cost less. If everyone paid it would barely cost anything at all.

The alternative is to have your search paid for by someone else, in which case all the incentives are aligned towards tracking and manipulating you for their benefit and away from providing decent search results for your benefit.

The marginal cost per user for a service like this is tiny, but there's a huge fixed cost to running the whole thing. If few people use it, it will cost a lot.

I agree that it's not good for things to split into premium paid services only affordable by the well off and rubbish free services for everyone else. However the solution isn't to abandon building services like this with funding models that correctly align incentives. It's for there to be a culture change towards understanding and valuing those well aligned incentives. This will bring more users, more services, more competition and much lower costs for users.


> If more people would pay it would cost less. If everyone paid it would barely cost anything at all.

Are you assuming it or do you know for a fact? That's normally not how startups work.. as long there is an increasing paying demand for their product, I don't see why they are going to reduce their price, quite the opposite actually.


If it were culturally accepted that search is something you need to pay for then yes, I am extremely confident it would come to cost very, very little per user. A startup with a small user base may well grow the way you say, but in a mature industry providing a commodity service to most of the population, margins will be squeezed razor thin and the prices will reflect the actual cost to provide.

Note that I'm not saying this is what will happen. I replied to a comment that was more about how we should want things to turn out. I'm saying we shouldn't denounce customer-paid startups for charging what they need to succeed in traditionally ad-funded industries, even if that means they start out as premium services for those willing and able to pay. Rather, we should cheer them on and advocate for the benefits their funding model brings, in the hope that they will succeed, grow, become commoditised and benefit everyone for pennies.

The alternative is to stick with the ad-driven, dross-filled panopticon we already have.


> The alternative is to stick with the ad-driven, dross-filled panopticon we already have.

Sure, I'm all for paying for something if they don't track me, if I can use their API as I need, etc.. the only thing that wasn't clear to me was the conclusion of "more people paying, cheaper it will get" because that probably isn't the case.


20 bucks per month is way too much, i like kagi but there isn't a justification for this price over free alternatives like duckduckgo


Try it free then ask your boss to pay it.

For me at least it saves me whole minutes and lots of stress on about half my queries which means at $20 it pays for itself in half a month and as a free bonus I get better results for the rest of the month as well, an ad free experience, no tracking and the opportunity to support a Google competitor.


I agree that 20 bucks is relatively a lot of money. On the flipside, running a search engine does not come for free, curious what the costs of Google would be without the ads.


I would happily pay $20/month to remove the massive frustrations I've had with modern search engines.


This is a fair point, high quality services should be available to all. That being said, the idea of paid search is quite novel and you need to target early adopters first. In my mind, this service is priced for early adopters and I only hope that the business is profitable from the very beginning. For me what WhatsApp used to charge for their service (1$ a year) is the epitome of high quality service for the masses. I would be very happy to see a service like Kagi scale to that with time.


> I have to say I support a lot of these initiatives, but I can't help but think altogether they're creating a division of the web into premium services the rich can afford, and low quality services for the masses

What do you mean? Like cars, houses, restaurants, groceries and almost literally everything else on the planet? :)


I really like the approach I've seen at https://draculatheme.com when you go to the purchase page of their PRO offering.

> Hey! You're coming from Some country where this could be too expensive.

> I believe in Purchasing Parity Power, and I want to make this affordable.

> If you need it, use the code SOMECODE for an extra XX% off the regular price.

I'm fine with paying, but it should respect purchasing power in each country. And I'd even go that far and offer the service free in some really poor countries.


Not sure I found what you mentioned on the page you linked to


Perhaps you are viewing it from a country, that has purchasing power this pricing was designed to. I'm viewing it from Czech Republic and see the message there.

There was a small discussion on HN previously about Dracula theme and this exact aspect: https://news.ycombinator.com/item?id=29495524


Note that their pricing discussions are rather open in their discord. They use both Google and Bing API for searches, which isn’t cheap (they said about $12/1000 queries).


Yeah, I am not sure if this will work, you can always install an adblocker and use Google or some bing based stuff instead, what is the value proposition exactly? I don't think their search results are any better honestly.


Well, services accessible to as many people as possible? Google, Facebook, Instagram? Is there any possible way to do that without using advertisement as the business model?


In US, something like this can exist but usually will be called communist.


I've used Kagi (and you.com) a bit, and I'm amazed at how much better the results were for the few queries I made. It wasn't blogspam any more, I got actual articles from actual sites. Hopefully I'll keep using them, but Google is overdue a disruption.


What do you mean by actual article? How is it different from a blog post? And why don't you like blog posts? Just curious :)


I meant I got high-quality blog posts from real people rather than copied/mass generated/spammy content, sorry.


> Be snappy and responsive on every device even without JavaScript enabled.

Huh, I thought I'd catch them out by disabling all javascript but they actually didn't use any for this webpage. Nice.


Not just that but all of Kagi Search works without JavaScript! We see JavaScript as a way to enhance product experience, not create it.


> We see JavaScript as a way to enhance product experience, not create it.

What an incredible one-liner, along multiple value axes.


omg they let me apply from/to dates on search results, I'm in love

just got my invite after a couple days of signing up, will be trying it out, but so far no stackoverflow/github issue clones, and they allow setting "preferred" and "muted" domains, I might pay whatever they ask

I'm keen to know their business plan however. Hope they don't get acquihired for the algorithms, never to offer these features again...


Yeah I wonder why google doesn't allow to blacklist domain for users. They could basically use it for ranking (eg. include paying, long-lived accounts to prevent abuse). Especially on youtube. They only have blacklisting in news but it's too dumb (eg. I want to block gossip but it's about sport person so instead it thinks I'm not interested in sport).

Wait I know why, because somewhere they have retarded KPI that will go down if people stop clicking on ads from spam websites that only show ads. Who would imagine


To be fair, Google also lets you filter search results by time ranges. The muted domains sound awesome, though!


If I tell you you just need to put a - in front of a domain on Google to mute it will you also pay me whatever I want?


Try to mute this way 10-20-100 domains in the same time, preferably on mobile. True, you can copy-paste this litany, but the whole process would be a misery.


I keep a running list of blogspam/SEO sites and use uBlock Origin's custom filters to remove them from Google/DuckDuckGo.

ie. google.*##.g:has(a[href*="realpython.com"])

duckduckgo.*##.results > div:has(a[href*="realpython.com"])


I just wonder why Google has never offered me to never show x domain again.


Probably because many of those domains are the same who contains most of their ads?

The rest is a bit tongue-in-cheek but I also can't think of better explanations:

Maybe because letting people block useless sites automatically means useful sites end up on top and giving the best results on top would mean users only stay on Google for 2 seconds and don't come back for that particular query while showing a mix of useless and ad laden non sense is pure brilliance:

- user need to stay longer on front page to decide which result to click => more ad exposure

- user goes to ad laden site => more ad exposure

- user comes back to try another result => more ad exposure

- user tries another etc etc

The whole thing seems to be a balancing act of how extremely bad they can make this slot machine before people revolt.

That said I'm open to alternative explanations:

- maybe they aren't that greedy, everybody smart just left as kt became harder and harder for non-SJW and those who are left either don't have time or are unable to fix it and it currently happens to work extremely well economically in the short run so nobody looks into it?

- or maybe as someone mentioned the other day everyone's search results are adjusted by an end-to-end optimizing AI which gets a digital dopamine hit everything an ad is shown or clicked and all good Google engineers have been struggling for years to get it back under control without breaking the cash cow in rhe process?


I like this line of thinking, that there’s something up with the paperclip optimizer. Google has always insisted that search engineering is different from ad auctions but I don’t buy it. If there’s a KPI for “dwell time” on the sponsored search results that ads up as you come back and try a different link, that would dive bomb the usefulness of the page.


Looks like a great product and I've signed up. One thing from your FAQ Page: You list Kagi as using significantly less CO2 than Google and Bing. Since Kagi is scraping both Google+Bing - wouldn't the Kagi CO2 total of each search be Google+Bing+Other Sources+Kagi?


To clarify: The CO2 measure is in the FAQ is based only on the amount of data transferred over the internet and Kagi wins here because we have the smallest payload. It does not include energy used to generate it, but Google claims it is carbon neutral since 2008 I think? and Bing is on its way to do so.


Test driving for a week now. Fast. Relevant results, haven't reached for alternatives (backups) just yet.

Disclaimer : haven't done a deep project yet ( new software stack or deep bug fix for example ).

Would pay for a neutral search engine, right now ( and I have no subscriptions atm, not even NF ).


If this really lets me block certain websites for good, then I'm sold.

I absolutely hate those sites that scrape other well-known sites (like StackOverflow, Quora, Reddit) and add ads and SEO spam. They are increasingly better at being on the top of the search results, and I cannot get rid of them.


I'm in the same boat. Right now I do this rather manually, e.g. uBlacklist on chrome.


While they claim they support privacy now, I'm still hesitant to create an account there. I can even use Google to search semi anonymously, given a proper ad-blocker, but creating an account to use a search engine? So I need to be fully logged in and identifiable and all my queries are associated to my accounts without any doubt because I signed in?

No thanks. I don't want that some hidden ToS change or future acquisition expose all my searches in the future, even if current search engine operates in a good fate.


Founder replied to me about the exact same issue in an older thread. https://news.ycombinator.com/item?id=29672560

I suppose it's the fault of online payment infrastructure more than them. Sure, they could do it the Mullvad way and accept envelopes with cash and an UID but that probably wouldnt scale.


I just got the invite yesterday and used it only two times. No seo spam at first glance, the first result is usable. Keep up the good work!


Another happy user here:

The time I save by not having to sift through all the utter nonsense that Google, Bing and even DuckDuckGo showes in my face any given day will easily be worth $10 a month alone and on top it is sanity preserving, privacy respecting and has no ads.

Its kind of like a polished version of https://search.marginalia.nu just with a much larger index.

The best explanation I can give is maybe:

Kagi is to other search engines today what Google was to other search engines 20 years ago: Outstanding quality, better than anything else I have tried - and rapidly improving - no nonsense (again like old Google) and nifty extra features (again like old Google).

(Note: I haven't tested Neeva yet so I cannot say if Kagi is better than them. But compared to all the rest I have tried the last ten years it is in a differen league.)


Gosh, I was not expecting my old shareware payment provider’s domain to end up in a new service such as this.

Makes me feel old.


People pay for VPNs but continue using the web with their normal accounts, thinking their privacy is protected. A private search engine would be a much better way to protect the privacy.


I also recommend to think about something like Nigma[1] (Russian metasearch engine) had done before, something along the lines of Wolfram Alpha - mathematics, physics, chemistry. It's possible to integrate with something like Sage Math for this purpose, should not be hard but would be immensely useful for many.

[1] https://ru.m.wikipedia.org/wiki/%D0%9D%D0%B8%D0%B3%D0%BC%D0%...


Is this Kagi in any way related to the defunct payment processor which used to bear the same URL?

https://tidbits.com/2016/08/04/kagi-shuts-down-after-falling...


Doesn’t seem like it. I didn’t see any reference to it on the website, either, but the last time this appeared on HN I was hoping someone would ask.

Since I haven’t seen the co-founder mention it, I hope he might make a little Easter egg or add it to the FAQ somewhere to pay homage to Kagi. That company made me my first money as a teenager with software.


Good i was getting so sick of non-premium search engines. Premiumness is what I have been looking for.


Thank you for this!

I would really love to pay you for this service. Chomping at the bit. However, I and many others would be much more willing to do this if we don't have to rely on your privacy policy (i.e., pinky swear) to keep our data private.

I think the simplest thing for you to do would be to implement a an option that allows paying with cryptocurrency (ideally Monero), and which doesn't require an email address. Just generate a subscriber number or username, allow login with user + pass (+ TOTP/HOTP) and allow me to fill my account with crypto!

Someone below mentions a token-based scheme for injecting noise; that's more elegant in some ways but of course more complicated to implement.

thank you for your work!


I wonder how it collects the results from “the best search engines on the market”. Bing’s API in Azure has very specific terms of use that do not allow something like Kagi, and Google Search API is not designed to build a generic search engine.


Somebody asked the other day.

Kagi pay for api access to both Google and Bing.


I'm actually super surprised Google does this kind of deals. I've thought they don't do this, becuase it would help competition to get started. Once you get traction, then it might be possible to drop the Google access.

This is what Kagi says in the FAQ.

"Our searching includes anonymized requests to traditional search indexes like Google and Bing as well as vertical sources like Wikipedia and DeepL or other APIs. We also have our own non-commercial index (Teclis), news index (TinyGem), and an AI for instant answers." [1]

[1] https://kagi.com/faq


I'm guessing Google does not this kind of deals unless very rare exceptions like Ecosia. You have some companies parsing Google results, like SerpApi, but it is a bit shady.


Interesting. How does one can access such APIs ? Do you need to know someone with an important position at Google ?


Contact me if you want to try this: https://www.mojeek.com/services/search/web-search-api/


No, I don't think so. Try searching: https://kagi.com/search?q=google+search+api or similar in your preferred search engine.


Have you tried to click on the results ? The public google search API is very limited, you can't do much with it.


The public api is limited.

Kagi pays.

I think anyone can pay for access unless they live in Iran, Cuba or North Korea.


Can you please link me to the API that you are thinking about ?


Last I looked into it was probably before 2017 at a time when Google insisted I was maybe a bot probably because I used very narrow searches.

I don't have time now but try searching with your preferred search engine or ask someone else.


I couldn't find the API you are thinking about. I don't think it's public and it would require a specific deal with Google.


We are an alternative: https://www.mojeek.com/services/search/web-search-api/


Both Kagi search and Orion browser sounds like a dream come true. I hope to get into beta soon.

Also wish for Orion browser to come to android aswell


Wish you well on your new venture. The domain name worked well for me for 25 years. Here’s to your future decades. Kee Nethery


Sounds awesome, I'd gladly pay for a product like this - I signed up for the waitlist :) Cheers devs!


Same here, I can't wait for an alternative with actually useful search results and good filters. Search is one product I use often every day but I'm stuck with either privacy-respecting options that don't yield useful results, Bing-themes or Google, which used to produce decent results but increasingly is just full of search spam. Good luck with this, I'm happy to pay!


Been using it for a week now, very happy with it. I still had a few times where google had better results but I will pay for kagi (so long as they have flexible payment options)

Check out their Orion browser too, been playing around with it.


My biggest issue with our current search engines is the lack of transparent open governance. My municipal water supply and the Debian project provide examples of better models.


Signed up! Good luck!

The logo is a g which is in the Google ballpark. why not K?


Good question!

The designer insisted that Kagi logo is "anchored" in the letter "g" and we had to agree. Besides, can't let Google own letter "g" too, can we? ;)


Stick it to the man! Good answer :)


Heh that was the first thing i thought, not a fan of the icon. If you claim to be better why use small mind tricks like this?


Can you show us some examples of where you are better. Pick something that normally spews out seo garbage like “Seattle gardening” and show me what you can do.


Kägi is also the name of a Swiss chocolate maker: https://www.kaegi.com/


I must say that going from google, DDG, Quant I'm most satisfied with you.com. This seems to be a similar engine.


When the second item on your startup's FAQ is how to pronounce the company name...


Happy user here, but has anyone tested kagi vs devonagent?


premium browser using our harvested data..

hypocrisy lvl 10000

> Kagi includes data from Apple, Google, Microsoft, Yelp, Wikipedia and others

come back after the reset Murica, this way of dealing with things is kinda outdated

is this yet another nonsense YC pump&dump thing?




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: