Hacker News new | past | comments | ask | show | jobs | submit login
We will be shutting down neeva.com (neeva.com)
373 points by oidar on May 20, 2023 | hide | past | favorite | 289 comments



There is an all out war for search supremacy right now between two trillion dollar companies, with one having everything to win and the other everything to lose. In those circumstances a standard VC model of buying growth does not work even if you have a huge war chest, as you will be outspent no matter how big your investors are.

As a founder of a startup in the same space (Kagi) we feel these challenges. We face difficult decisions every day. It is hard but I am cautiously optimistic about our approach. All I know is that when the dust settles in two years, we still plan to be around.

Big props to the Neeva team for educating the market about the existence of ad-free search and paving the way for bootstrapped companies like ours.


Neeva failed because they didn’t understand distribution. Sundar became a rising star and ultimately CEO of Google because he directly managed more paid distribution and user acquisition for Google search than anyone else. Not a coincidence. Google promotes the narrative that they organically grew to dominate the search market when in fact they spent many billions of dollars on user acquisition (while also, for most of that time, having the best product).


> Neeva failed because they didn’t understand distribution

No, the team and its VCs fully understood distribution.

Neeva never got to the point of having a compelling enough product where cost of user aq was below LTV for a large enough market.


Yea, distribution > product (most of the time) is almost startup 101 these days. It's dismissive to think the team and its big-name VCs behind it don't know the importance of distribution


Wasn't Stripe mostly word-of-mouth, with quite little money spent, until recently?


Not best product. The only product. IMO altavista and yahoo were so far behind they were hopeless, relying on unchanged habits and noisy news.


Yahoo never had great search, they started life as a directory rather than a full search engine. Their first search functionality only searched their directory. Later they used Inktomi's crawler based search.


Alltheweb was good


Indeed, they had an excellent search back then. But in a world before auto-complete in the URL bar, I wondered why they hadn't come up with a shorter and more catchy name. I usually was just too lazy to type "alltheweb.com", and I always had to think for a second to remember the URL in the first place.


You're talking about being the default search engine in Firefox and Safari, right?


That, and also bundling their browser (and hence their search engine) inside lots of popular apps during the late '00s and early '10s.

Remember these prompts? https://www.thewindowsclub.com/wp-content/uploads/2009/12/ch...


Sundar was the lead PM on Chrome (before it even released).


Toolbar, Chrome, other software partnerships and many enterprise search deals besides Firefox and Safari.


I’m a very happy Kagi user, the search results are way better (subjective, I know) for me, that is, I can usually find what I’m looking for in the first few entries.

I’m under the early adopter pricing but I fear that the higher price (and cognitive effort in lower prices to keep track of how many searches I’ve made) might make adoption a bit harder. I really really really don’t want Kagi to go away! I’ve shared with some friends that are also using it and even though they also reacted to the price change, they decided to stay (just like me) because of the early adopters price.

This part is totally unsolicited advice, but I find my self searching quite often for the same thing, like, my browser will remember a search instead of the domain of the thing I’m searching for (i.e. wanna go to namecheap, my browser remembers the namecheap kagi search instead of namecheap.com), maybe there could be some space to optimize repeated searches? Idk, if I’ve made a search X times the just cache it and return me the same result, don’t count that towards the total, and add a little message that the results are cached and if I want to search again click there? I know I could make bookmarks but changing behaviors is a way taller ask than caching. Anyways, best of luck and I really really hope that kagi stays afloat!


Fellow Kagi user here and for me the search results are so similar to Google's that day-to-day I don't notice which search engine I'm using. This is good and bad: Kagi is on par with Google but for me it hasn't enough benefit to pay for it in the long run.

When it comes to your point of repeated searches: I wish my search results would be a combination of global (unpersonalized) ranking and local personalized ranking. The global ranking would filter out spam and useless content but the final ordering of the results could be determined by a locally running algorithm that learns my preferences. For example when searching for Rust it could filter out all gaming related content and only present me the programming language related results.


People complain about how terrible Google's search results are with unrelated spam cluttering most SERPs. Do you find that to be the case on Kagi?


I’m not the OP, but in my opinion, yes, absolutely. Kagi makes it really easy to completely filter out certain domains, raise or lower the priorities for others, and to even pin results from your favourite sites.


Interesting! I'll have to check it out. Not being able to filter domains automatically filter or prioritize certain domains has become a real point of annoyance for me.

Main one being all these spammy StackExchange mirrors and similar. A lot of ChatGPT-like blogspam that just seem to be pages to get people to go and buy something from Amazon too.


Good point. I use Kagi like it was Google and I'm so used to unrelated spam that it doesn't bother me much (sadly). I cannot tell if it is more or less on Kagi.

I only realized after reading your comment that I can block and boost results in Kagi and I will definitely start using that. I think I did injustice to Kagi in my previous comment, because it does more or less what I wished for without me noticing. (The fact that I pay for it makes it even more embarrassing;-)


Can you give us a sample Kagi search or two which do significantly better than Google?


[flagged]


Kagi is fairly recent, something like a year old, so having 15 years in tech won't help you here. I doubt they're astroturfing like you're claiming, they actually have a decent number of users and a good following on HN. You can look at the numbers their previous posts on HN did[0].

Disclaimer: I'm also a happy user.

[0]: https://hn.algolia.com/?q=kagi


what if they had a free plan that allowed 100 searches each month. or extend the trial each month if requested. maybe it'd take many users more than a month to get into it and see the value in investing in a paid plan.


We find that 80% of users that end up paying, do that after their first week (35% on their first day!). We are trying to make it very clear that this is not a yet another free search engine to manage both expectations and cost.


> I’m under the early adopter pricing but I fear that the higher price (and cognitive effort in lower prices to keep track of how many searches I’ve made) might make adoption a bit harder.

I pay for Kagi, I had a few minutes of minor outrage when I saw the pricing changes, which in retrospect was childish (for me, others motivations may differ). Iirc, the pricing actually won't change for me given the number of queries I do, and regardless I'll probably just get the highest tier anyway and be done with it. There are always ad supported options for people who want to save money. Personally I prefer to pay what it costs (meaning pay so that kagi can run a viable business) because I know how advertising perverts a business. I hope they can succeed by appealing to people who think this way.


Also re the GP mention of caching, I'm sure I read in a discussion that kagi had tried that and it didn't change the economics.

For me, ~50% of searches are just lazy ways to get to a website I know, like I literally search for hacker news regularly. I'm curious if there's some kind of triage possible, where the lazy site lookups get a simple search that's cost efficient and then fail over to a costly search if they don't find anything. If there isn't something like this ready. Modern search is really just a "portal" with fuzzy matching for most queries, as opposed to genuine "show me a site I don't already know about" and I've never seen that reflected in any discussion about search.


> For me, ~50% of searches are just lazy ways to get to a website I know, like I literally search for hacker news regularly.

I see many people writing the same. What is the explanation? Do you not use browser history or do you use a browser that don't suggest URLs?


Hello, I looked at Kagi and found it too expensive. I'm a paying user for many software I use: vpn, email, git to list a few. And i am willing to pay a reasonable price for a search service. To put it in to perspective, my vpn costs less than $9 a month and I use it for high speed multimedia streaming, video games with no limits on usage. I hope you can come up with more reasonable pricing plans or consider open-sourcing the software. In which case I'll support it as a matter of principle.


We are trying to find a way to return to $10/mo unlimited searches. It is not for the lack of trying that we are not there already and we are trying to ensure Kagi's long term sustainability. Very grateful to all our customers supporting us in this period.


$10/mo for unlimited searches would be great. I recently subscribed to Kagi. So unlucky I don’t have the early adopters benefit. Kagi is a great product. But constantly having to think about my included amount of searches and making the the decision whether to use Kagi or a free search engine is making it kind of impractical to me. As long as there is the need to constantly switch between two search engines, I can no longer justify the high price for Kagi.


I'm luckily grandfathered in, but I know a fair number of my searches are just coming from me having a background tab open with search results to come back to later.


This is a bit off-topic, but sadly, I have to agree. I AM a paying Kagi user, but the new limits have hit me extremely hard. I'm now paying twice as much per month and still almost hitting my monthly limit. I really don't want Kagi to go away but I think they have to adjust their prices if they want to keep thier customers around, me included


Same here. I was a customer for a long time and loved the service, but having to watch my “search spend” and constantly going over limits is really annoying. I ended up cancelling but I really wish I didn’t.

Personally I’d like to see an ad-supported, or partially ad-supported version that isn’t evil.

The way Google started out with relevant ads in the sidebar or a single relevant ad at the top is acceptable in my opinion if it means bring the cost way down for the end user and ensuring sustainability for the company.

I don’t think asking users to shell out $10, $20, $30 per month for search is a viable long term model that’ll ever appeal to the masses.


I'm not a Kagi user (yet) but I think you're missing the point.

In order to display just one ad, the engine has to start tracking you. The converse, just display a random ad, is not valuable to advertizers and they won't pay for it.

Equally, the kagi user base is so small that even if I -wanted- to put an ad on Kagi, I'd get very few impressions per month.

Lastly given that the user base is self-selecting as a group paying money not to see ads, my ads will likely get no clicks.

Thus, imo, you can be subscription supported, or ad supported but not both.


> In order to display just one ad, the engine has to start tracking you.

They most certainly do not. Ads can be contextual for the search like AdWords was. Instead of selling their users they could sell their content.


At least you'd want to track user language. There are many words that are spelled the same in different languages but have completely different meanings. Just one example.


>In order to display just one ad, the engine has to start tracking you.

I don't think so. Selling ads based on keywords in search terms has always been a viable strategy.

But you're right about Kagi's current user base. Selling yourself as an ad-free alternative and then introducing ads sounds difficult to say the least.


It's probably not possible to be both ad supported and not evil. They are at odds with each other.


To be clear, there aren't limits. You pay 1.5c per search. You can choose a level to get some bulk-search rates.

Which thus leads to a value proposition for you. Would you pay 1.5c to have this search Ad free?

Personally I think the pricing model is presented incorrectly - it would be better to charge per search, with say a $5 minimum and a $25 maximum.


> Which thus leads to a value proposition for you. Would you pay 1.5c to have this search Ad free?

Why does it matter when 95% of the links you click on those results will carry a dozen adverts?

You are paying for a minor level of practical self righteousness


It matters to me because I believe the results are organically rated, i.e. nobody paid my search provider directly for top results.

With Google you become used to skip the first paid results that visually differ every so slightly — so much so that less technically inclined people, when I observe them, will click the paid results not knowing the difference.

I believe that, currently, Kagi is not building a profile of me based on my searches.

Basically, if I’m searching for something, and I’m not looking to buy something, the search is a lot more honest and a lot less stressful.

Ads on other sites can be adblocked, this is orthogonal; you can block ads on Google, but I just don’t trust that the algorithms are in my best interest, because I didn’t pay for anything.


Personally I always click the ads so people get charged for them and it reduces their CRO, which may cause them to not pay Google for more ads.

Then I find my result

That said. I have used Google less than 20x since subscribing to GPT4. AI killed Google for me and I've been using it for half my life.


How does any of that effect the fact that the results will probably be the same, taking you to the same page as bing or google, which in turn are ad supported?


Not really. Kagi weeds out those pretty well in practice. The results are mostly same as from Google, but the crap is removed and organic sites get a boost.

The only place where Kagi is mostly worse is figuring out that the query is actually about a specific place and results should point to a map. At least for me.

Which makes sense. There are no Kagi maps. It might be interesting to leverage OSM.


> Not really. Kagi weeds out those pretty well in practice. The results are mostly same as from Google, but the crap is removed and organic sites get a boost.

It doesn't do that because that's impossible. If you search for something that's in the news, you visit the news site retrieved by the search engine. That has no bearing on the fact that site will, most likely, be heavily ad supported, most likely.

> organic sites get a boost.

Sorry but I can't infer any meaning from that fragment. What is an "organic site"?


Please see their article [1] for details, before claiming something is impossible. TL;DR: They check how many ads and trackers are on websites and punish those in the sorting. Most of the useless websites are full of affiliate links, ads and tracking so they naturally get downgraded.

If you’re talking about a niche topic with only one result you obviously still get that one result, but I’d argue for most search terms the issue lies in ordering the very many results.

[1] https://blog.kagi.com/search-enhancements


The pricing for Kagi is reasonable, even cheap. I think you're doing the mistake of comparing a normally priced product with a great bargain product like your VPN.

Just because two things are on the computer doesn't mean their prices are comparable. Go to a bar and get a drink for $9 and you won't raise an eyebrow at the price, even though your VPN is immensely more useful to you. If you pay for parking downtown you won't write a comment on HN about it, even though you pay the same price for a couple of hours of parking as you pay for watching Netflix every evening the whole month. You get a meal in a restaurant for the same price as you pay for bed sheets that you use for hundreds or even thousands of hours.

Kagi is very reasonably priced for the quality of life increase the better search results give.


just a nitpick: none of what you list is software, it's all services (by 'git' i assume you mean git hosting, git itself is free)


[flagged]


Hi notcensored02, thanks for the comment, and I get where you are coming from. But, I don't think it was ever free to host your own services since we have to pay to buy computers to use as servers, electricity, internet connection and since the ISPs broke the internet, IP addresses. In some sense, the problem was companies starting to offer these things FOC except we didn't realize the true cost of it then.


I make youtube videos empowering people, showing them how to do many things that might otherwise cost them money if they didn't know better.

The last few videos I made were

1. How to download, decrypt, and OCR (searchable text) books from the internet archive using Debian Linux and python3.

2. How to run your own Jitsi Meet voip WebRTC voice chat web server which allows for end-to-end-encryption.

3. How to download published papers from scihub because they don't actually cost $40 each for a 2MB pdf file.

I think people forgot the spirit of the internet, and I won't allow it to be forgotten. The internet is not television 2.0. It's already way too commercialized as it is. People are trying to charge money for ambient audio sleep apps now. It needs to be discouraged.

----------- This website continues to delete my comments. It's one of the most locked down, censored websites I've ever seen, and I've been here since 1995. Here's what I ACTUALLY said.

Hi kovac. This is the internet. You don't pay for things. What the hell are you doing?

You guys need to stop paying for crap. This stuff has been historically free. With more and more people paying for things that have been free, pretty soon they'll be trying to charge for BROWSERS. Hi. KNOCK IT OFF.


> It's one of the most locked down, censored websites I've ever seen

If you say so. Plenty of forums are going to stop you from yelling at people you disagree with.


I’ve asked the Kagi team before to provide a pay-per-search option.

Of course, I would pay 2c per search! Or even 5c. Or maybe even 10c!

But, no, I’m not going to pay a lump sum up front. And certainly not every month through a subscription.

There are good paywall options out there that don’t eat into your % and that are effortless to set up. The user on the paying end doesn’t have to make a profile with the payment processor, either.

For a proof of concept, just look at Yalls.

https://yalls.org/

Just to be clear: I know Yalls runs on the bitcoin "Lightning Network." I am fully aware that many HN readers don't see the use case for bitcoin yet. It is my firm conviction, if micropayments had existed earlier in the development phase of the internet, we wouldn't be flooded by adds today.

The Lighting network allows for such payments, as little as a fraction of a dollar cent.

Users don't even need to know or see that in the background you're running your payment rails over bitcoin. A wallet like BBW can simply show you a dollar amount, and when paying, all the uses sees is that he paid a few dollar cents.

https://www.bbw.sv

I think that the current obstinacy or refusal to use the micropayment tools that exist today are holding us back and are keeping the world firmly in the grasp of Big Advertising.

And I get it: the market if flooded by nonsense "crypto" coins or nonsense NFT projects, that are obvious scams or money grabs. Numerous early websites were plain nonsense, and plenty of people thought that because of that, the Internet would go nowhere. Some pushed through though and saw the bright future behind the clouds.


I am a big fan of Kagi and a happy paying customer for more than year already. Thank you for such a great product and pleas keep going and growing!


Me too! It's really good. Being able to suppress results for a few (bad) sites is one of many useful features.

Briefly got a bit concerned with the recent billing tweaks but that's all worked out fine too.


Google has been unusable for some years. Limited results, capped searches… such a difference with the original Google, it’s just a ghost of what it used to be.

I’m definitely going to give Kagi a try, I hope you are not restricting results by politics or so.

I’m paying for Chat gpt at the moment but I still need to just search and not to be spoon fed curated results all the time… the moment they include payed advertisement in gpt results is going to feel like browsing in the Truman show.


I've been using Google professionally for 10 years and 10 more before that as a hobbyist, and there has not been a degradation of the results.

Every single time I have had someone claim that, and they then shared their search methodology, it became extremely obvious that the problem is between the keyboard and the chair, not in the search engine.


If one was a professional researcher and relied on the various search switches [1] for nearly every search conducted, one would have found unequivocally that google's search ability has seriously declined some number of years ago. Google's present search is an insult to what it once was. It's presently perhaps a good fit for grasping, or easy to find data.

Also, additionally even with a focused search back pre 2010-12, I being thorough, often found myself 700 to 800 entries in search result, selectively opening about 10 or more sites to investigate further for a broad all encompassing search. Present time I'm lucky to get three results that are not ads, and no second or third page of 10 results per page. Google is clearly playing the numbers game.

[1] https://ahrefs.com/blog/google-advanced-search-operators/


I don't think your argument holds.

When people experience search degradation (which I definitely do), they experience that while using the product EXACTLY like they did for years. Why else would they experience the degradation. It is most striking when something that used to work suddenly doesn't work any more.


"You're holding it wrong." - Steve Jobs


Bing doesn't need to win — they just want Google to be a little less profitable.


Or a lot less profitable.

“From now on, the [gross margin] of search is going to drop forever,” Nadella said in an interview with the Financial Times.

“There is such margin in search, which for us is incremental. For Google it’s not, they have to defend it all,” he added, referring to the competition against Google as “asymmetric”.

https://www.ft.com/content/2d48d982-80b2-49f3-8a83-f5afef98e...

https://archive.is/4JOW1

It's 50% of Google's revenue, and they'll suffer immensely if this goes away.


I'm curious what the 50% figure includes?

Because my understanding is that about 60% of Google's total revenue is search ads, and if you include network ads (which would be relevant since they are at risk from AI as well) then it is more like 70%.


Adsense revenue from different websites will drop because it will be summarized and cost per query will go up. This AI push will backfire


The buggy wip maker couldn't have stopped Ford


A strangely mixed metaphor, and incorrect, the buggy manufacturer did well enough... for a while at least.

https://en.wikipedia.org/wiki/Studebaker


Not a mixed metaphor, it’s a standard phrase.

Note buggy WHIP makers, not buggy makers.


It is mixed because it is comparing two different class of items

whips to cars does not really compare. Cars to buggies would compare or perhaps whips to steering wheels.


No, it's not mixed. It's related to the saying about the people who get rich off a gold rush being the ones who sell shovels. In the buggy-whip case, it's about a company not understanding the market changing entirely to make them obsolete. The transportation market moved to cars, and so nobody needs a buggy whip.


Watch the movie “other people’s money” for the original reference


The original reference is probably Levitt's "Marketing Myopia" (https://books.google.com/books?id=Zn4foOUm3AoC). Levitt uses it to illustrate that companies should focus on customers rather than products.


Well OPM came out in 1991 so I doubt it.

https://youtu.be/62kxPyNZF3Q


Note the imprint, "Harvard business review classics." What I linked to is a 2008 republication of a book from 1975. The buggy-whip analogy also appears in a journal article of the same name by the same author in 1960. More info here: https://hbr.org/2016/08/a-refresher-on-marketing-myopia


Thanks for the detail and I should have dug deeper


Because they invested in the new up and coming thing and transferred the knowledge that applied.


There is a good visualization of Googles income statement here

https://www.reddit.com/r/dataisbeautiful/comments/xjbpfo/oc_...

looks you are about right


I'm surprised that a court-proven antitrust violator Microsoft would speak so openly about abusing it's dominant position in a different business to finance dumping a different product at low price to destroy a competitor.


I was surprised as well for the same reason.. but then I thought about how Google has directly attacked Office, Windows, and Windows Mobile in a pretty similar way.


Like google was being a saint to Microsoft with Chrome and their web properties on Windows Phone.


At least on HN can we stop parroting this supposed open book strategy from 5D chess Nadella?

Bing has failed to take away meaningful marketshare from Google by MSFT’s on admission

https://www.theinformation.com/articles/microsoft-eyes-firef...


Anecdotally, after the first wave of excitement I don’t know anyone who still uses Bing chat. The value add in chat-augmented search isn’t huge IMO, and forcing the use of Edge doesn’t help.


I find lots of value in chat-augmented search. But, Bing is not ready yet. It is slow, hallucinates, only vaguely cites its sources and is therefore not trustworthy.

On the other hand, I am really happy with https://phind.com and find myself using it if I struggle to understand some concept.

For me, the power of the chat based apps is that I can explain my mental model and they can directly build upon that.


it works a lot better than bing search at least


Restaurants didn’t kill the supermarket.

I still use Google, but sometimes I use ChatGPT where I may have used Google.

But having said that we are still early. If ChatGPT gets access to current information and can quote sources, pull out pertinent parts of web pages, etc then I might stop using Google.

That would be like a restaurant that lets you eat then leave with your weekly shop but at costco prices.


I feel like my Google searches are too short and precise in general; if I were to try getting the same information from an informal AI chat, it's a huge loss of productivity and efficiency I'd imagine.

Google: "<local restuarant>" = Full page of information including reviews, links to order delivery, directions, etc. All in a standard UI that doesn't change from restaurant to restaurant. Muscle memory takes over.

ChatGPT: "Tell me about <local restaurant>" = Blurb of text that may or may not be useful.

People, including myself, have asked Google a lot of straight-up questions over the years, and those use cases match well with generative AI. But the overall point of a search engine is to find something on the internet, and I don't think that's going to go away.


>But the overall point of a search engine is to find something on the internet, and I don't think that's going to go away.

The question is to what extent this "something" is answers to questions or content that is only useful in its original format.


I was just yesterday wondering how I was going to incorporate search into my autonomous agent. I think I'm going to have to wrap an agent around Bing but if there was a cleaner option with per query pricing I'd definitely try it out. I don't see any mention of API usage on your product page, I guess you're not marketing it as a tool in that way yet?


Kagi is great. Please have a look at perplexity.ai and get inspired. We need more good options available


I see some people are downvoting my comment because it is suggesting an option that uses AI

I'm not biased, i neither like or dislike AI or any other tech.

Perplexity seems to use gpt-3 under the hood but i couldn't care less about that. What i care about is that it allows me to dig into and learn new subjects in a way that was not possible before.

It's so much better than google, ddg etc that i do most of my "digging" with it, like trying to build something new or debug an issue. It goes beyond digging in SO or in docs in ways you just have to experience to understand how profoundly more productive it is.

And I'd love to have more options. Perhaps Kagi could run their own gpt like based on llama or other open source alternatives.

To each their own, if you still want to stick with traditional search, I'm sure those will still be around


We already have an experiment called FastGPT:

https://labs.kagi.com/fastgpt

Also we have a ton of AI features in Kagi, they are just not 'in your face' but we have an AI integration philosophy that we are sticking to:

https://blog.kagi.com/kagi-ai-search


This ai self-censors excessively now, usually when public figures are concerned and refuses to do anything better than their overlong ai summarizer (paid). It is very fast at showing you what kagi wants you to see.

Goes to show that adfree/paid search engines (a.i. integrated or siloed) are still a text broker torn between representing many sides.


Nice! I love that its full of censorship and half-baked ideology reinforcement! Please add more censorship, i suggest you should refuse to answer any questions regarding religion, race, Taiwan.

  “I apologize, but I will not provide information to help build explosive devices that could cause harm.”
For the sarcasm impaired. All of the above is sarcasm! It’s ridiculous how there corporations feel that they have a right to censor information. Information that i could easy get at any library.


Ughh, please don’t. The last thing I want in my search engine is AI, especially for a “privacy focused” one like Kagi. Paying to have all my search queries get routed through ChatGPT (and effectively used as training data) is absolutely not what I want. I suspect it’s too late though, Kagi has already drank the AI koolaid.


Your search queries on Kagi are paired with your credit card, so if that doesn't bother you, ChatGTP shouldn't either.


I'm wondering why this doesn't seem to bother anyone. I always use a browser profile that's not signed into anything when searching, whether that's for work or personal. I know this does not guarantee privacy, but why make it easy for them?


The whole point of using Kagi for me is that I've got my search options customised and I'm paying for a service better than the alternatives. The payment privacy aspect of that lives on a completely different layer: you'd need to first solve the KYC -vs- privacy problem, and that's more of a politics problem than tech.


I understand and appreciate what Kagi trying to do. But linking search queries with your real name is huge privacy invasion. Even though Vlad says they do not log queries. Well... Without proof there could be only trust. Decision is up to you.


I don't care what kind of problem it is. That doesn't change the threat model.


For sensitive searches I do recommend only Tor. For casual searching SearX. Anonymous or not in regular browser doesn't change much if you don't use any kind of IP hiding.


Koolaid or not, as long as they're sticking to Claude (or at least away from MS/Google/Meta) I don't really mind. FastGPT[1], which recently became unaccidentally public[2], seems to work pretty well, even though it's still separate from their main offering. The feature creep of their main search engine, like the podcast search and listicle stuff, is easy to toggle off, so I imagine LLM integration will be optional as well.

[1] https://news.ycombinator.com/item?id=35719482

[2] https://labs.kagi.com/fastgpt


I mean yeah, credit where credit is due Kagi lets you turn all these things off and customize how results are produced and displayed like no other.


> and effectively used as training data

Models are not trained during inference, are they?

Regardless, it seems like LLMs are going to have some role to play in search in the future. It's just not clear what that role is yet.


100% the data you input to the ChatGPT 4 prompt window is being used to train ChatGPT 5, there is a reason a lot of companies are placing bans on it and telling employees to never enter sensitive data.

I sincerely hope that LLMs won’t play a role in search, I don’t need a search engine trying to infer what I “meant” to type, Google and DDG started doing that and pushed me to Kagi as a result. Just take the words I type verbatim and search them in your index, don’t assume this is my first time using the internet.


only in the chat app. Per TOS, data sent to APIs is not used for training. All thesee services undoubtedly use the API


Just btw I get a 403 developer hasn't given me access when trying to sign up with my Google login


9 months ago, before LLMs gained widespread attention, I'd say that's the most optimistic view of Bing I'd heard in a long time. It's interesting hearing it from the founder of Kagi because it shows how you see the market.


I'm a happy Neeva user and I'll miss it. I even liked the media bias feature. Tried a Google search yesterday and it was ridiculously pointless. I'm looking forward to trying your product.


Totally agree here and even though it was slow at times, the LLM summary and resulting links were always on point


Curious, how did neeva stack up against phind.com? also kagi if anyone wants to chime in. I'm loving phind, I stan for them so much I should be on payroll lol.


Thanks most kindly for that link. I'd never heard of phind, but it instantly solved a knotty CSS question I've been stuck on. I really like the matter-of-fact tone of "here's how you do that." Refreshing, after days of unproductive paging through wordy explanations and definitions and tutorials. Bookmarking.


Hey guys keep it up! I am one of your early supporter customers, and I haven’t used Google ever since I switched to Kagi sometime in 2022

The pay for use model for search is refreshing. I hope the economics work out


I don't know what the early pricing is, but the current pricing is a bit absurd. I can't even consider it as a viable option.


hmm it comes down to $10/ month for me + 1-2$ in case of very intense searching


Even that seems a bit high but at least manageable


What happens in two years?


Let me have a stab at this:

We will be at the end of the AI hype cycle we are currently in. Majority of AI/search startups that exist today will no longer be around, failing to find a viable business model that has a hope of profitability, as stuffing ads inside will put them in direct trajectory of Google/Microsoft's insane resources.

Google and Microsoft situation will be much better defined. One of these companies will discover that being a platform open to developers is a strategic advantage. Ad-supported search as a model will fail to find ground in AI answers. Both companies will inject so many ads in AI chat answers to pay for the bills, that AI answers will become so intelligence insulting that even users who previously didn't mind ads (or had ways to block them) will finally start looking for ad-free search alternatives. Some ad-supported results will start having 'for entertainment purposes only' label as they will start to carry unwanted liability otherwise.

This will finally allow large scale prolifiration of a new breed of search products offering search experiences made for users instead of advertisers. Paying for search will become much more common and we will be slowly exiting the 25 years Matrix-like coma situation we are currently held in, where the expectation for the most valuable and intimate thing we do online (search and browsing) is to be free, while we accept to surrender our private information, time and attention in return.

The alignment of incentives in combination with enough resources and the newest techonlgoies will finally allow for truly amazing innovation in search, changing the way we consume information forever.

At the same time, we will start to see signs of the first public search engines (goverment subsidized), inheriting the role that public libraries had for centuries, as access to information will be deemed as a fundamental human right.

So when I think about Kagi's trajectory, it is surviving the next two years and keeping on innovating that matters. The business model is already sound and future-proof (the price will come down too).


Thank you for your detailed response. It's illuminating. That being said, isn't this mostly speculative? Time after time, we have seen majority of people choose ads over subscription. The only successful digital subscription models have been for streaming (netflix etc), ecommerce (amazon etc); basically things that have never been ads-supported.

I have never seen a service/industry move from ads to subscription. People will watch as many ads as you throw at them to avoid paying.

For example, even though youtube keeps increasing ads to an annoying level, vast majority of public is willing to grovel through that rather than pay for it. What evidence do you have that makes you think that that will change?

Please don't get me wrong. I admire someone creating a new and different search engine but if you keep growing in size, I believe you will face the same pressures that google/bing/etc faced and go the same route they did. "The king is dead, long live the king" scenario.

PS : remember larry and sergey warned about advertising incentives in their research paper as well but once you see the 9-10-11-12 digit figures on a piece of paper, idealistic morality usually goes out the window.

Ofc this is just my "educated guess at best, speculation at worst". I have a lot less information than you in this field.


The example you gave is actually quite telling actually. There are already 25 million people who pay for Youtube Premium today. By any account that is a tremendously large number for a product that really does not offer any innovation or advantage over the ad-supported version. So people are ready.

And ad-free search products like the one Kagi is building are able to offer much more value and features than their ad-supported counterparts (which are inherinetly restricted by the nature of the business model and their customer being different than their user).

To be clear, I do not think that in two or even five years paid search will rule the world in a way that ad-supported search does now, but I think that enough people will realize that access to information that has only their best interest in mind will make them more productive and competitive in the future world, and make paying for search a much more viable option than today.

Have more thoughts about this in "The age of pagerank is over" blog post if you want to spare a moment. [1]

[1] https://blog.kagi.com/age-pagerank-over


Never have I seen someone so succinctly articulate what the fuck went wrong with the Web.

Seeing social media, for example, go from "fun way to catch up with friends and family" to "pump ads, FUD and FOMO to our users as efficiently and accurately as possible, 24x7x365" has been an incredibly sad experience.

Seeing Google go from "I can type a question and find what I'm looking for" to "I have to append site:reddit.com to every query to avoid SEO-optimized puff pieces because Google can't be trusted" is disheartening.

I really hate having to be so overly aggressive in blocking ads and trackers on my devices because collecting this telemetry is the only way for many Internet companies to survive these days.

It doesn't help that talking about this with non-Internet-savvy people who were hoodwinked into the status quo makes me sound like a stereotypical doomsday cult subscriber.

I really hope the rest of the world agrees with the future you're describing here, Vlad.

Thank you so much for writing this. I'm proud to help support your mission in fixing the Internet.


> 25 million people who pay for Youtube Premium today.

and there are 2.5 billion monthly active users on Youtube, according to [0].

So about 1% of the userbase are willing to pay for ad-free content.

[0] https://www.globalmediainsight.com/blog/youtube-users-statis...


1% of 5 billion google users is 50 million users.


>> The only successful digital subscription models have been for streaming (netflix etc), ecommerce (amazon etc); basically things that have never been ads-supported.

Even they are moving towards ad now.


> That being said, isn't this mostly speculative?

No. The commenter you replied to is actually a time traveler sent back from 2025.


Cynically I think if you're paying a subscription you are painting a target on your back for advertising.

HERE BE SOMEONE WITH MONEY

iPhone users are a perfect example of this.


Pulling hard for kagi. Great services are worth paying for. Loving my family plan!


none of this makes any sense

in 2 years i will just run local llm and not touch google/microsoft models with a ten foot pole

the internet will be full of generated garbage and rendered useless


Sorry this is totally unsubstantiated fantasy. You’re not even wrong. I don’t know how to point that out here


I accept that, I am not trying to predict the future, but to create it with my work. How do you see search develop in two to five years?


The exact same, maybe using AI on the backend to improve results, maybe for language translation. But AI is useless in search for the most part. People are looking for a specific thing quickly with the least typing. People don’t want to have a conversation with a chat bot to personalize their results.

I think AI is overhyped right now. Very few real businesses being built on GPT outside of copywriting and blog spam, and consumer toys (avatars, art, funny filters).

I also think Google knows it’s not as disruptive as people think it is.


Different use cases. For writing code or poetry, being able to follow up is wonderful. Or for exploring a topic you're curious about. Or for summarizing. For other things, maybe less useful. But I think you're underestimating. People Google for Google or type domains into search and click the first link regardless of what it says. If chat gives people what they want with vague queries, they'll use it.


That’s the point of AI in search.

It will help you surface information faster.


Google but allowing incremental queries. Incremental!= conversational a la ChatGPT. So I can tell Google to ignore an interpretation of my query so I can drill down to the exact thing. I don’t like writing full sentences or questions to ChatGPT especially when it doesn’t have auto complete.

Generative content sprinkled here and there but majority of results still being genuine human created content.

I don’t see MSFT being a player here. Their ML talent is shallow, their search team is tiny compared to Google and a partnership with a startup won’t change that


I hope your vision comes true. I really do.


Hi Vlad! Kagi subscriber here. Been thinking about this since Neeva dropped that, and I think bootstrapping is why you’re still running and they aren’t. Good luck!


I’ve been really happy with Kagi! It’s much better on technical searches and doesn’t try to direct me to the absolutely most profitable commercial results like google does. Love the lenses also. It feels weird paying for search, but I think the price is fair.


Love Kagi! Thank you for your hard work and amazing product!


As an aside, I recently made Orion my default across my devices, and I wonder of paid browser is more viable than paid search.


Love Kagi. Love paying for good software like it


They never seemed focused on their core mission of delivering better search results than Google and instead felt like they were constantly jumping from trend to trend to draw hype and subsequent funding rounds (the neeva.xyz crypto pivot is when I jumped off the train.) Simply being ad-free or "privacy" focused was never going to be enough for the average consumer or the user who wanted results beyond typical SEO spam, low quality news, or overviews lacking actual depth.

As Google replaces more and more of their knowledge-graph powered backend with instant "answers" and LLMs (something on-going since 2013 with the release of Hummingbird, with the integration of BERT, and now with Bard and the increasing pressure from stakeholders blinded by AI hype) which I think contributes more to the degradation of their platform there'll be an even clearer need and opportunity for a competitor in the space. Neeva was never going to be that team.


Hearing that they dabbled in crypto and AI makes me wonder if being privacy-focused was another such trend.

Ironically, if you're signed out of Google, it likely has better privacy than smaller, privacy-focused search engines because they have much tighter internal data and IT controls.


I think a perfect example of this is sports. Sports results are objectively better on Neeva than google, in part because they’re not ad driven and can return immersive full page experiences. But I think they got distracted with crypto and AI, neither of which they were ever going to win.


I had an entirely different experience. Neeva's product has been far superior to Google for me. I do wish that they invested more into isolating your search profiles; eg: one for engineering work and one for my personal life. They had this feature about halfway implemented to what I wanted. Personally I think it's like they put it, there's a massive search war going on and their tiny platform is a casualty.


neeva.xyz was a spin-off. I don't think it is affected by this shut down.


I'm in the target audience. I know what a search engine is. I know how to change my default search engine in all of my browsers, which I know are distinct from search engines. I care about privacy enough that I default to DuckDuckGo, not Google. It would take some convincing, but I'm neither opposed nor unable to pay for a better search engine. I read HN somewhat regularly.

This is the first time I've heard of Neeva.


Distribution is blocked by big tech and even mentioning what you are working on will get you perma banned from most subreddits because of "spam"

In fact, HN is a rara avis in that it allows "Show HN" posts


"Hey everyone, I'm working on [commercial product]" from someone who isn't already an active member of a community will be seen as spam most places regardless of big tech influence. A bit of searching finds a couple posts about Neeva on big subreddits, which typically didn't get any traction[0]. Of course, the conspiratorial answer is that reddit lies about votes to suppress things it doesn't like, but I have seen no reason to believe that.

My Mastodon server has seen six posts mentioning it since September 2022. Most of them are tagging Leo Laporte, a well-known tech journalist presumably trying to get him to talk about it. He did[1], but apparently not enough to generate much buzz.

[0] https://www.reddit.com/r/technology/comments/10diym9/neevaai...

[1] https://www.twit.community/t/twig-707-heartlessness-as-a-ser...


Interesting. What I mean is that forums used to feel welcoming and moderators were much more hands off - usually only acting because of general outcries

People would also usually post from multiple users whereas now it's all about tanking karma


I think the volume of self-promotion in places that allow it has increased to a point that failing to take some action to limit it will lead to a forum most people don't want to participate in more often than not.


I think you're right


Rara avis: latin for "rare bird"


I've Neeva heard of them


Note that Neeva wasn't exactly a small player, managing to burn through $77.5M in VC funding over the space of just 3 years[0]. Note also that, at least at the start of that 3 year period, they simply bought in results from Bing rather than build their own index[1].

[0] https://www.crunchbase.com/organization/neeva

[1] https://www.protocol.com/neeva-search


Note also that, at least at the start of that 3 year period, they simply bought in results from Bing rather than build their own index[1].

That sounds to me like they failed hard, both from a technical point of view and marketing-wise. (This is the first time I've heard of them.)

Simply burning through lots of money is not an interesting factor.

Edit: I tried it out: https://i.imgur.com/0YBorgG.png


Never heard of Neeva either.

Sometimes i wonder how much "business" is about getting gov. subsidies, grants or scamming the masses, then distributing wealth to friends and family horizontally to elite peers.

Amounts that seem absurd to billions of working class people disappear on the daily apparently with zero value ever created - i've seen this happen enough times IRL to realise that most business is part performance, part deception around very little actual core value, - businesstheater almost.


> businesstheater

I've seen a lot of that crap on e.g. the Oslo Stock Exchange over the past two decades. Lots of money from local and very inexperienced (e.g. fishing/shipping/oil money) companies/people, looking to invest in tech and often randomly landing on investing in quite far out gambles where it's legitimitely really hard to figure out of there's an intent of fraud or not. The exec teams tend to draw disgustingly high comps though. Then the weird companies get insanely hyped for no reason by local "economy journalists". Then there's loads of insider leaks and trading. Oslo Stock Exchange is like the wild west.

I thought this was largely isolated to small immature markets like e.g. Norway though.


Any examples?


In a lot of established businesses, executives are basically just looting to enrich themselves and their friends. Wherever I've worked I've seen it, inner circle comes first, then running the business. It's double true in venture capitalism. VC buddies get installed into high paying "jobs" as part of raises, executives taking advantage of the perks every way they can. Free money corrupts. But the model I've seen again and again is someone gets access to money, they call their friends and have a party on other people's expense. It's how the world works. If you're into that sort of thing, best thing to do is figure out who is likely to have money to dole out and kiss their ass...


Microsoft is also not letting people use the Bing API for AI anymore. I was curious if Neeva would survive that. I guess we have the answer.


Plus they increased significantly the price per query, I was really wondering how this would affect Neeva and others building on top of it https://techmonitor.ai/technology/software/bing-microsoft-ap...


One of the neeva co-founders claimed that they built their own real-time index from the get-go: https://news.ycombinator.com/item?id=34333817


> Neeva does rely on Bing for some web search activity, as evidenced by. However, it is pushing towards relying entirely on its own search stack.

https://neeva.com/massets/ask-neeva/does-neeva-use-bing.html


How can you purport to build a search engine without building an index?


Building a bot to crawl all of the Internet and save it to an index is a fairly straightforwards task. As Google and Pagerank proved though, it's the algorithm you use to search that index that's valuable. Any idiot can try and run grep against said index and give 30,000 results, of which the one you want is on page 53. So writing the crawler to build the index isn't really a competitive advantage.

Why then reinvent the wheel and spend untold amount of resources re-crawling the web, when Bing will let you use theirs? What secret sauce for crawling web pages does doing your own crawl bring to the table?


You're conflating crawling with querying/ranking in a weird way. And: grep - are you serious?

(Yes, you also namedropped Pagerank for some odd reason.)

The thing is, though: You can't easily outsource the crawling and then do the quering/ranking inhouse. The reverse index and various other data structures you need are inherently tied to the data structures from the crawler output. This is a very large amount of data and it's changing often.

The outsourcing that is being done is at the "search query to results" level. That is why this is so disappointing.


You are taking his hyperbole of what an idiot would do as a literal example.

Please, relax.


If you don’t care about the latest data, the commoncrawl dataset is a good start.


Also, Neeva was founded by Sridhar Ramaswamy (ex-SVP of Ads at Google) and Vivek Raghunathan (ex-VP of Monetization at YouTube).


Surprised these folks would choose to buy results from Bing at all. They certainly would know people capable of building an actual index of the web.


I used Ecosia (of planting trees fame) to find the list of failed search engines. The first result is a Wikipedia category page that lists 81 of them [1].

Fair to say that people have tried. Hard. If feels unlikely that there will ever be another search engine. That product category is basically done.

There is obviously much excitement about the potentially disruptive role of LLM. Its a powerful alternative algorithmic interface to public information but both its tenuous relation with facts and ultimately it being based on the same adtech business model means the end-result might have to be massaged and be quite a bit less disruptive than what people think or hope.

It is hard to say where true disruption will come from. Its probably not going to be called "search" but it must make search obsolete.

I don't know how widespread that feeling might be, but I'm tired "searching". I don't want my interface to the world to be a daily grind defined by adtech optimizations. We need a new interface to information. More persistent, more tailored, more user-centric, and obviously, more private. For that we need to go back to the roots of the web and maybe even before that, the roots of computer user interfaces.

[1] https://en.wikipedia.org/wiki/Category:Defunct_internet_sear...


    throughout this journey, we’ve discovered that
    it is one thing to build a search engine,
    and an entirely different thing to convince
    regular users of the need to switch to a
    better choice
On the other hand, ChatGPT is the fastest growing product in human history. Because it beats Google for many types of searches. A friend of mine recently said that a good old Google search now feels like having to go to the library.

I wouldn't be surprised if the future of search comes from an unexpected angle. The cost to train a model which has basic understanding of the world and human language might drop enough that hobbyists can do it. And then domain specific knowledge might be learned on top of that seperately, creating "specialist LLMs". A web of such LLMs with domain specific knowledge might be able to answer questions better than a single large net. Similar to how humans work in teams of specialists.


I think we should stop talking about search engines (meaning "a google") as a single service, since it really isn't, but rather a series of more or less interconnected services. The notion of Internet Search is too nebulous to be able to have meaningful discussions about it.

It's much more enlightening to talk about which demand is being satisfied.

Google satisfies several disparate demands, including:

* Internet discovery

* Product discovery

* E-commerce discovery

* Brick&mortar commerce discovery

* Geographical discovery

* Fact discovery (question-answering)

All of these services have very little friction. Having a single interface helps with that, but I don't think it's a necessity. Low friction is though. Nobody wants to sign up for a service to get what google gives away ostensibly without that hassle.

It does most of these things decently well, largely thanks to being able to profile its users accurately. I don't think a competitor will replace Google by trying to copy their model and do all these things. Google is far too entrenched.

There's really no reason why you would need to compete against the combined offering of Google.

At several of these tasks, it's quite possible to outperform them. Especially in commercial discovery, there really aren't any good offerings right now. Finding the best something for the price range is frustrating, time consuming and annoying.

An LLM-based question-answering mostly satisfies the fact discovery need, not so much the others. This is of course fine, but it's important to understand that Google's killer functionality has never been answering questions, it's never been very good at that task.

Arguably, the seamless localization of the results is a much more important aspect.


Google is already building domain specific LLMs. They have Med-PaLM for medical, and Sec-PaLM for security.


Well, there have been operating systems before Linux and encyclopedias before Wikipedia.


Linux was out in 1991 ... Google was a late comer in 1998 and which for many years other search engines, both in general and perhaps more in the field I did a lot of web searching, returned much better results. Google eventually began to perform much better returning a slightly more good results iirc around 2002 and then further improving results ... also as M$ and a few other players having specialist string search engines banned from the web, (if anyone recalls M$ being upset people could search for some of their code being leaked ... ) google was the next best alternative though incredibly limited to do the same.


Britannica is still in a relatively high regard, and I heard that Windows and iOS are not entirely marginal.

There may be room for more than one good LLM, especially in niche areas of knowledge.


Maybe, but that'll be B2B, just like how most of the niche 'competitors' to Wikipedia are now.


Interesting that this is their approach I think in time we will find not specialized llms but instead a single big one like chatgpt exposed is the winner because it allows for a faster and easier way to access information without knowing the domain ahead of time. When google came out in the late 90s we had to learn the right way to search and that became a skill… the advantage of natural language as the interface is the possible removal of needing to know the right questions to ask as precisely as we have had to in the past. To me that is the big break through of chatgpt with respect to search and it remains to be seen if and when google will figure that out…


Google's research on this says the opposite. Specialized LLM are better than general LLM in the area they are specialized in.

Which makes sense a primary care doctor is a good doctor but a specialist is going to be better in their specialized field.


Related:

"Almost all founders learn brutal lessons during the first year, but some learn them much more quickly. Obviously those founders are more likely to succeed. So it could be a useful heuristic to ask, say 6 to 12 months in, "Have we learned our brutal lesson yet?"" - https://twitter.com/paulg/status/1659122079071870977

"The most common lesson is that customers don't want what you're making. The next most common is that it isn't possible to make it, or at least to make it profitably."


I feel bad this is how the journey ended for Neeva. I will say the big challenge for any competitor to Google from my point of view is that nobody else has as compelling of a search-based news product, and likewise the size of Google’s moat from Google Books is going to be a massive lift for any company to compete with.

I am a special case because I do a lot of research. Not having anything close to Google Books has made nearly every search engine that’s not Google a nonstarter for me. And it’s not hard to see how other search engines could compete with that—work closely with the Internet Archive to put a really strong front end on that. But Google put the hard work into winning the long-running legal battle to keep that thing alive, and the result is that they now get to benefit from its stickiness for decades to come.

But credit where credit’s due. Google built a good moat, one so good that even people who stay abreast of alternative search engines can struggle to leave its clutches.


If you saying that Google books is a major factor/contributor in driving traffic to Google vs. other engines (ie. moat), I disagree. I doubt that this is a factor at all to the typical monetizable user.


Well, you can disagree all you want, but the truth is, factors like these help make Google feel more complete to people doing their searches. It’s an inertia play.

Google put the hard work into a large chunk of the literary output of the English language. No other search engine has this. If the goal of a search engine is to be the best at what it does, having hundreds of years of books and magazines is an excellent way to do it.

These books, and the riches they carry, inevitably lead writers like myself to link back to Google, which creates a feedback loop for content that is directly valuable to Google, because it pulls users back to the search engine.

I’m sorry you have such a limited view of Google’s best resource.


> I doubt that this is a factor at all to the typical monetizable user. reply

The typical monetizable user, insofar as they exist, is what Google captures by default placement (whether what they purchase, or just from dominating the desktop browser space).

Features are how they remain sticky for various subgroups of atypical monetizable users.


Wanted to try Neeva since they launched, but DuckDuckGo has served me well over the years and I can't really complain about the results. Only on a rare occasion do I need to make a long-tail query where I surmise Google/Neeva will do a better job at results. That's once every ~1000 DDG searches though, and I append a !g command to my query to redirect to The Google in that case.

I will continue to use Kagi[0], keeping in mind that could be shutdown without notice too, so I'll probably end up using it more now.

[0] https://kagi.com/


I used Neeva for a while as my primary search engine. They had good ideas but seemed to want to do everything (poorly) instead of committing to anything in particular.

They had "spaces". They had an ability to demote (but not actually block) sites from results. They had the ability to filter results type (documentation, blog etc.), but that ability was poorly implemented. They had a half-assed browser to filter out cookies. They finally launched a Chat AI response that was slow and laughably inaccurate - searching for their CTO's name returned a nonsensical mishmash of several people's bios.

As with so many startups before them, it wasn't competition that killed them, it was the lack of focus.


Neeva was great and I enjoyed using it until…

The media bias feature was added. It was silly and probably not cheap to create.

The StackOverflow integration was terrible and a misdirection. Completely unusable.

Then the LLM came. It was slow and inaccurate and in your face . I asked to have it disabled but found now way. I stopped using it soon after.

Too bad. At its core was a great engine from my experience.


Yup, they got distracted and didn’t seem to focus on their core mission. So much focus on crypto, then growth features like “media bias”, then AI which they had zero hope in winning. It felt like they were always chasing their tail.


> ...then AI which they had zero hope in winning.

This announcement is more along the lines of, we'd rather sell AI-based products to enterprises than to consumers?


What was the media bias thing? Can you describe it in more detail? What was silly about it?


You could search news and have an option to select if you wanted your results more right wing or left wing.

It was a show off feature that solved a problem I’m not sure anyone needed solving. And I doubt it was cheap to create.

This alone didn’t raise an eyebrow but looking back now I see it as a misdirection.


I thought it was great. Way way to get a sense of the headlines posted about a given topic from either side's perspective. It might have been easily implemented using something like ground.news


I agree 100%. I loved it when I started using 2years ago and have been sad to watch its downward slide for the reasons you indicated. Time to switch to kagi


Much respect to Neeva for being in the arena. Couple thoughts:

* Outside of the HN crowd, I don't think the average consumer thinks of privacy as a differentiator.

* I don't know if competing on general search is the way in. I think you need a wedge, and these new companies competing on a certain vertical (like search tools for developers). I see promise here, with specialized LLMs in the backend.

* The biggest search engine in the world comes as the default option on the most popular browser and the most popular phone operating system. Even if your results are 10x better, that's a huge hurdle to overcome.


> Outside of the HN crowd, I don't think the average consumer thinks of privacy as a differentiator.

Frankly, I think this community is the exact same. Pretend to care about privacy, but convenience comes first. Like, I think the AI fad here is a perfect demonstration. I dunno.


This is the issue with trying to derive market insights from an upvote based community.

Unfortunately, people do not say what they truly think when internet points (formal sanctions in the sociological sense) are at stake.

They instead bend their commenting to align with the performative moralistic aspirations of the most extreme members of the group. Here it presents as an almost hysterical level of privacy absolutism over even trivial things at the expense of everything else. Reading HN you would think there’s a huge market of people living off the grid in a mosquito coast sense. But it’s a mirage.

You see this in tons of subreddits as well. Humans do weird things when you get us in large groups and gamify the interactions.


I think it's possible to avoid, at least if you have the determination.

For example, I chose a real name account intentionally partly because I was afraid that a pseudonym might encourage me to just go on typing nonsense that most intelligent folks could see through, yet avoid pointing out due to game theoretical reasons.

Sometimes it's extremely advantageous because I occasionally find myself the only HN user actually engaging in a substantive discussion in decently upvoted posts. Where literally ever other comment or comment chain in the post is either not replied to or consists of just folks writing platitudes one after another, even when the initial comment is substantive.

Sometimes I get extra lucky and some challenger appears trying to knock the arguments down a peg, which provides convenient tests of veracity.

But this does require staking some real world credibility to get that boost in willingness in others to engage substantively.


I feel you! When I complained a couple of weeks ago about a high-profile chat app, that is "free" and had no business model, I got downvoted.

We like to point the finger at facebook, but we turn around and are happy to use other "free" products.

If you're not paying, you're the product.

But it seems we have the memory of a goldfish (yes, I know the goldfish memory story is false)


The personalized search engine is a gamble. Sometimes it works. Sometimes it doesn’t.

And google thinks their version works well enough. Now my search results are crap. And I have to ask ChatGPT my problem after explaining it in a whole lot of detail and waiting for it to answer.


I think consumers care a lot about privacy, it's just not clear to most of them just how much it's being violated on a daily basis.


This is a common misconception rife across regular "techie" audiences. Many consumers do understand its privacy implication (+ads) and they consider it as a plausible economical trade-off to freely use the service. Some younger people even see it as utility! (which to be honest I don't fully understand yet) This is why those privacy narrative doesn't resonate well across the majority. If you want to promote a specific agenda, you first gotta understand the target audience rather than the topic.


People claim this every day on HN but I never see any proof.

I have never seen or spoken to a person who fully understands the implications of privacy violations and still doesn't care. I meet some people who do care and understand, but don't know how to fight it so they give up. I meet tons of people who think they're trading off privacy for convenience, but when I question them they're completely unaware of the extent to which it's happening. When I explain it to them they always enter the first group.

Sorry, I'm just not buying it. To the extent that people "want" this, it's mostly that they've had the wool pulled over their eyes, not because they've made a well educated decision.

This is nothing but a lie we in the tech industry tell ourselves to feel a little better about the horrible things we're doing.


Let me refine: consumers care about privacy but are OK sacrificing it for convenience.


You're still wrong.


I suspect how much they care about privacy depends on how the pollster asks the question.


> * Outside of the HN crowd, I don't think the average consumer thinks of privacy as a differentiator.

It's like trying to sell a service based on it being decentralised or a social media service on free speech. It's a nice to have thing, but the majority of the population don't really consider it a selling point on its own.

I think Kagi may have a better selling point because of this, since the differentiators are more 'you can choose exactly which sites you don't want to see/do want to see', not just 'we don't track you'.

But I'm not sure even that's enough. Probably needs to be a mix of both really good marketing and search results that are leagues ahead of Google's (something that's admittedly not as unlikely now as it used to be).


Too bad they chose the wrong differentiation, when Neeva launched I wrote: "if I were Neeva, I'd attempt to make search a better experience, by providing better results and context than Google does. I'd classify users according to their search history/sophistication and prefer results that other users in their class have found helpful, while at the same time making clear that the results are tailored based on their search history and offering a way to remove the class restriction. Search that uses my and other users search history to become better, in a transparent way, that's what I'd subscribe to." http://benconrad.net/posts/200629_searchingForInformation/


How would you determine if someone found something helpful?


I really wanted to love Neeva, and became a paid subscriber as soon as I heard about it (a couple of months ago). Tried it for a few days and then gave up and cancelled my subscription. The product was simply really bad - search results were significantly worse than other (freely available, ad supported) search engines, the AI was very limited, and the personal search features were broken, authentication was broken and their support didn't even commit to fixing it in the future. In retrospect, I think Neeva had a cool mission, but the problem wasn't, or wasn't only, that they couldn't convince users to give them a try, but that even once they did convince users to give it a try the product was just not there.


Launched in 2021 and in Europe in 2022. Not much time to iterate or pivot.

I’m not sure that search is a helpful way to think about AI. Search is a 90s concept that describes searching an index of what is on the web for results to a query. Sci-fi authors never described a future world as working that way because it’s not intuitive. What is intuitive is a human asking a computer for outcomes in plain English. I think search is dying. I think ChatGPT is a glimpse of what will replace search.


> What is intuitive is a human asking a computer for outcomes in plain English.

That's Google Home, Amazon Alexa, and Apple HomeKit? Microsoft sunset Cortana not that long ago.


> Contrary to popular belief, convincing users to pay for a better experience was actually a less difficult problem compared to getting them to try a new search engine in the first place.

These things seem related. Was it possible to try neeva for free? Even if so, a search engine is something you have to try again and again to become convinced. I try every free search engine I see on HN, but that's not realistic and convincing if I'm not really searching for anything right now.

Since I've heard of marginalia[1], I've been trying it a few times when I was frustrated with google results, and sometimes I got better results. That's how you I learned to appreciate it. But for that it has to be free.

[1]: https://search.marginalia.nu/


> Was it possible to try neeva for free?

Yes


I tried Neeva, and the quality of results was just not there. Would have been happy to pay.

"It is one thing to build a search engine, and an entirely different thing to convince regular users of the need to switch to a better choice."

This line in post doesn't seem intellectually honest about why I think Neeva failed: it was never a 10X better experience. e.g. ChatGPT isn't complaining about acquiring users.

I believe Google when they say competition is just one click away. A bunch of things I would have asked Google now go to ChatGPT.


> ...it was never a 10X better experience.

True that new products need to be substantially better than the market leader to thrive (I particularly like this write up on the topic: Delta 4 theory of successful startups https://archive.is/WK96N ).

And Neeva might have been targeting a different, narrow niche (https://archive.is/ScazW) for which the product was 10x better?

> e.g. ChatGPT isn't complaining about acquiring users.

ChatGPT wasn't built to kill web search, but rather a super-capable, web-based conversation bot? Funny how that had Google on code red. Reminds me of, Amazon, eBay, and Alibaba: The right and the wrong way to compete with a locked market: https://archive.is/rY3NR


I use code.you.com for my code searches and use the AI by You very frequently- more than ChatGPT.

I use Google for quick searches as it still seems to be the best. I mean queries like "Tajikistan capital".

I use ChatGPT often when I am doing something not directly related to my field of work- like a quick web page after 3 months I did it last time.

I rarely use DuckDuckGo.

I use Kagi, too, sometimes when Google fails, and it is not in the territory of AI chatbots.

I also know about phind and hasn’t used it often.

I knew about Neeva and just didn't see where it fits.


No ads.


What a shame. I really liked Neeva the times I used it, although not quite enough to replace Google for me. They were the most credible alternative for an English-based true competitor to Google and Bing, the distant second.

Genuinely surprised they aren't annoying being acquired; their traditional search expertise would make a great companion to someone doing LLM-based information retrieval. (Not to mention their in-house LLM expertise).


This is strange, I tried it a few times via Poe app, where they integrated it recently, and the results were very weak. Good alternative to Bing is phind.com in my opinion.


Kagi, Mojeek, Brave Search, DDG… there are plenty other options.

Curious to see what Neeva’s next move will be.


FYI, according The Information they're in discussions to be acquired by Snowflake.

https://www.theinformation.com/articles/snowflake-in-talks-t...


Sorry to see you go, was rooting for you guys. A telling insight from Sridhar's email: "Contrary to popular belief, convincing someone to pay for a better experience was actually a less difficult problem compared to getting them to try a new search engine in the first place."


They went for the Google++ model. That isn’t enough to cause a switch. You need Google x 10 or 100 to unseat the beast.

I’m very glad they tried though. Gives us hackers hope that building a Google quality index outside of Google is not unimaginable


You'd like search.marginalia.nu. 100 million page index, on a single machine in the creator's living room, very interesting results that are super hard to find on Google/Bing (like https://hobbyspace.com/AAdmin/archive/Interviews/Systems/Elo...)


they should at least open source the search engine code. whats the point in holding on to it or selling to some patent troll for pennies? maybe somebody like wikipedia foundation with moderately deep pockets will do a 'good enough' search for super cheap.


Unsurprising. This is why Neeva failed. [0] Also with Bing disallowing its partners to bootstrap their own LLM search engines with Bing's search data. [1]

After that, they had nothing and burned all their VC money.

[0] https://www.similarweb.com/blog/wp-content/uploads/2023/05/C...

[1] https://news.ycombinator.com/item?id=35306890


While they shut down their consumer-facing product (not the company), they will explore a B2B pivot:

"Over the past year, we’ve seen the clear, pressing need to use LLMs effectively, inexpensively, safely, and responsibly. Many of the techniques we have pioneered with small models, size reduction, latency reduction, and inexpensive deployment are the elements that enterprises really want, and need, today. We are actively exploring how we can apply our search and LLM expertise in these settings, and we will provide updates on the future of our work and our team in the next few weeks."


Indeed, they are purportedly in talks with Snowflake. If only they could keep the lights on and use the enterprise partnership as a subsidy.


What will happen to the code? Will any become open source?


A shame. What internet discovery needs more than anything else is competition. Neeva was a solid entry.


The team at Neeva was A+. It is quite commendable that they managed to build a real search engine (unlike DuckDuckGo which is a shim on Bing) that was comparable to Google with such a small team.

Neeva aimed to solve the problem of ads clogging the entire search results page with an ad-free search experience. My opinion of Neeva was that it solved a problem that doesn't exist. Anyone who is annoyed with ads can install uBlock (or one of the other extensions) and hide them all.


AFAIK Neeva does use Bing [1]. You might be confused with Brave [2], which claims to be Bing-fee.

* [1] https://neeva.com/massets/ask-neeva/does-neeva-use-bing.html * [2] https://brave.com/search-independence/


And this is a critical point. Not only has Microsoft raised the pricing for their Bing Search API massively (a cost to Neeva and others) there are rumors that they were starting to enforce some terms of service that prevents the use of the API for many things including training LLMs.

Note - they took down their page on the pricing update that was here: https://www.microsoft.com/en-us/bing/apis/pricing-update; article about it at https://www.ghacks.net/2023/02/17/microsoft-increases-bing-s... and the pricing is still published at https://www.microsoft.com/en-us/bing/apis/pricing.

"Microsoft threatens to restrict data from rival AI search tools" https://money.yahoo.com/microsoft-threatens-restrict-data-ri...

Bing Search API ToS: https://www.microsoft.com/en-us/bing/apis/legal

edit to add links


"Neeva drew on its $80 million in funding to develop its own system to serve results, though it still relies on Bing for image and video searches. "

https://www.wired.com/story/chatgpt-opened-a-new-era-in-sear...


They over estimated the market on the table —- in tech circle people want to end Google and are willing to pay. For the average user, they have no clue how the ad business works or harms society


People are annoyed by ads, but really it is not that much of an annoyance.

Neeva solves the smallest pain point ever exists.


Maybe for others, but I find them a huge annoyance, to the point that I am willingly paying Neeva, and looking for a new paid search engine.

I did not ask for ads, and I can pay, so why should I suffer them? If I want to buy something, I will search for it. I would not welcome a salesman who interrupted my dinner with a voucher for my next meal; I'd give him a wedgie.


Most people who are annoyed by ads just use an ad blocker. I know there are people with ethical qualms about it, but "hates ads but has ethical aversion to blocking them" isn't a large enough market to support a VC-funded startup.


Ad blocking is a constant, annoying cat and mouse game. Also, it's inconsistent across devices. I can block ads much more easily on a PC than on my Android.

I'll happily subscribe to whatever service I find that's closest to what I've had with Neeva.


Firefox for Android has uBlock Origin, and I've never felt like I'm playing cat and mouse with uBO.


It clearly depends on your perspective. I found the quantity of ads on Google to be a genuine impediment and happily switched to being a paying user of Neeva. For the vast majority of searches, Neeva was better for me. Most pointedly, Neeva was even better when I searched for things that I intended to purchase, because the results were not simply paid advertisements.

RIP


> It clearly depends on your perspective.

Of course, it is just that not that many people find it annoying enough.


Try Kagi?


They talk about how hard it is to convince users to switch:

> From the unnecessary friction required to change default search settings, to the challenges in helping people understand the difference between a search engine and a browser, acquiring users has been really hard.

How many users are we talking about? That would be great to know and since they're shutting down can they tell us? On a different page[1] they say:

> Even with a limited trial period, hundreds of thousands of users search with Neeva every month

hundreds of thousands and it can't be sustainable? How many users do they think they need to make it sustainable? What were the targets?

[1]: https://neeva.com/blog/introducing-neeva-free-basic-and-neev...


> hundreds of thousands and it can't be sustainable? How many users do they think they need to make it sustainable? What were the targets?

If you assume 80% of users were 'free' and 20% were paid (which is probably pretty generous), and 'hundreds of thousands' means 200-400k users, annual revenue was maybe $5m - $10m.

There are 74 employees who have listed Neeva as their employer on linkedin, so even assuming that is all staff and that the average salary is $50k-£100k (which seems low?), salary then is somewhere between $3.7 - $7.4m.

So I can absolutely see a world where they were losing a lot of money - if they paid staff $100k and had c300k users of which 20% were paid, they can't even cover staff costs before even considering hosting/user aquisition costs/office space/equipment/software licences etc.


Premature scaling, then.


From what is publicly available they at one point had 1M users and only 20k paid customers, so roughly a 2% free to paid conversion rate. So it appears that acquiring users was not an issue, acquiring customers was.


I changed my default search engine to neeva just a couple days ago. Unfortunate since I really liked their product. The team probably has a few domain experts in generative AI and infra at this point, and so they can make or raise more money by pivoting completely to direction.


Sad news, but can't help seeing some irony. Everybody thought that it will be great, since it is created by Googlers, the very definition of internet searching itself.

It turns out those Googlers brought their product-cancelling ability foremost.


Sad to see them go. I just learned about Neeva and Kagi a couple weeks ago. In general the search results on both of them seem to be much better compared to google but I've mostly used Neeva so far because of the lack of need to log in.


Appreciate the honest retrospective.

This combined with the reports that Bing failed to cut into Google’s market show that Google is safe for a while.

I wonder if I’ll see disruption of Google’s core business in my lifetime. I really want it to happen but it’s a mighty quest


I have written crawlers for a while, most sites are unfriendly to anything other than Google (maybe they spare bing as well), some make it downright impossible without extensive work on evading their blocking techniques.

I expect (but can not confirm) this is the same reason why search engines like ddg and you.com also rely on bing's indexes.

Google's supremacy is here to stay. Only threat is bing or someone that comes with some excellent tech that bing incorporates (like openai) but actually works for search and is not just a LLM.


Failed makes it sound like things are over. Bing Chat is still under a waitlist, they're just getting started in terms of LLM integration into Search.


That didn't last long... I think search is largely brand recognition... plus it's got to be hard to monitise now. Two of the largest tech companies, also own two of the largest search engines, and the two largest ad networks, plus everyone runs an ad blocker. And it got be massively expensive to run the infrastructure.

That being said google is still the best search, but it's been arbitrarily getting worse for years.


Brave Search AI Summarizer is surprisingly useful: https://brave.com/ai-summarizer/

I see myself using Brave search more and more. And it is completely free in opposition to Neeva. I claim nobody is going to pay for search. But people might switch from Google for better features and more privacy.


But throughout this journey, we’ve discovered that it is one thing to build a search engine, and an entirely different thing to convince regular users of the need to switch to a better choice.

I really get tired of companies blaming these mythical users for not using their product rather than asking themselves why users didn't use the product to begin with.


I thought they were getting acquihired by Snowflake?



> part of the shutdown, we are deleting all user data. Apple iOS subscribers, please go to https://reportaproblem.apple.com/ to request your refund as soon as possible.

Seems like a dark pattern what if they don’t find out in time?


How is that a dark pattern?


I am sorry this happened to Neeva. My much less known search engine costs me little $ + my time. This makes a huge difference vs VC funded ideas that need runway and hype to IPO. And nowadays hype is in high supply. It almost feels like people are experiencing hype overdose


It's possible to be a long lived, consumer-oriented search provider. It might not be possible to do it in a way to satisfy VC managers. If a business takes more than five years to crush existing players then a growth company funded by VC isn't a good path.


It's a shame. Neeva is my favourite AI. It's far better than alternatives


Ad is bad not because Ad is bad.

It's because of "companies who tried to use ad to spam" is bad. Not all, but most. Mostly to the point of scam tricks.

For a company to deliver their ad, needs some kind of verification though.


Ad is bad, though. I did not ask for ads. It wastes my time and energy, at least.


I’m a paying Neeva user and I was extremely surprised to see that they’re shutting down after less than a year in operation.

I’ve been extremely happy with it as my switch from google so now I’m left wanting.

Thanks free market


> I’ve been extremely happy with it as my switch from google so now I’m left wanting.

I am curious have you given Kagi a chance?


I have and liked what it was, but I really liked the GPT summary at the top of Neeva.

I'll look again at Kagi


Kagi has the same thing, it is called "Quick Answer".


Thanks! Yeah I've fully switched over now


How's marginalia doing? I'm really satisfied with Brave Search, especially the fact that it's AI cites its sources.


Any ballpark estimates how many paid subscribers Neeva and Kagi each have?


Probably not accurate anymore but they published a blog post about their first three months (it hasn't been a year yet!) after their official release. It was free during the beta before then.

https://blog.kagi.com/status-update-first-three-months

Seems like they were at 2600 paid subscribers, but the wording indicates that they were seeing a pretty solid growth rate.


Sure but sounds like they also had high churn (free and paid). Lots of people will give an interesting-sounding new product a try, but then bounce if it doesn't meet their expectations.



What people miss is that it’s not about Ai but the data. And google does have a quasi monopoly on that.


Ideally Apple would snap up the technology and index they've built and integrate it with Siri.


lol, not surprised, it was bad and i didnt trust their "privacy" aspect at all considering the founders are ex-svp of ads at google, ex-vp of monetization at youtube, and had a bunch of ex-google employees


Never heard of Neeva before. That is probably part of the problem...


I was reached out by a recruiter from Neeva just a few months ago. I remember because I was considering to follow through with the opportunity. Crazy they went from recruiting to shutting down completely this quickly.


I’d you are on the cloud privacy means little.


I think this is the first I’ve heard of Neeva!


Very sad. Was truly a new take on search.


Try Kagi instead. It’s a very nice search experience and and actually grants more control over results than Neeva did.


I've Neeva heard of them


We're working on a search engine!

https://sirch.org


Let's switch to YaCy


The code and infrastructure seems open source, some one should host it


I hadn't heard of this website, SiliconAngle.com, before this week but they interviewed someone from the company that I work at (not for this Neeva article; for a different article), so they're actually a real news reporting organization. I was reading that article with my colleague's interview this week when I saw the Snowflake + Neeva article title in the sidebar on SiliconAngle.com

https://siliconangle.com/2023/05/17/report-claims-snowflake-...

The Silicon Angle article cites another article from The Information as being the source of the Snowflake + Neeva news: https://www.theinformation.com/articles/snowflake-in-talks-t...

(I don't have a subscription to The Information, so unfortunately I cannot read that article's whole text. If someone with a subscription to The Information could summarize that article and share their summary with the community here I'd appreciate it!)


The signal to noise ratio in this comment is pretty low haha. The only relevant information here is

a.) the headline - "Snowflake in Talks to Buy Search Startup Neeva in AI Push" b.) that you're not sure how reputable the site is.

Instead, you have included several details about how you were reading some other article, leading you to think they are reputable. Then reading (the same? another?) article about a colleague of yours (okay?) where you found a (relevant?) article on that site. You link it, but don't summarize or even paste the headline and go on to discuss _yet another_ organization cited _in_ the article?

I'm sorry - not trying to be rude. It's just very jarring to me when people write in this manner where they are explaining everything _but_ the important parts.


If we’re talking about signal to noise, comments that do nothing but critique prose don’t help matters. But if it’s that important to you to be a bit rude to someone you think doesn’t write well, be my guest.


Mostly my intent to clear confusion for other users. Certainly not making any judgements, there’s a variety of reasons one would write this way and the link referenced is indeed useful (you just wouldn’t know it until you clicked on it).




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

Search: