Non-AI unlimited plan

marc-eu

mcc111

"How about a more expensive non-AI plan?"

If there were a plan that removed all "AI" capabilities from Kagi, I would sign up for it immediately even if it cost more.

I switched from Google to Kagi exclusively to get away from Google's AI features. If I had the option, I would not be supporting companies that sell generative-AI products at all— I don't think such companies deserve my business. However there is no such option in the search engine space, so I am now using Kagi, because it at least allows me to (partially, awkwardly) turn the AI options off. Specifically I am currently using Kagi's custom CSS to hide all AI buttons so I do not activate them even by accident. But this is inadequate, because it means I cannot use the Kagi app (custom CSS works only on the website). Additionally, I don't get to use Kagi News because Kagi does not offer any option for viewing news stories except with AI "summaries". Kagi should offer a non-AI plan option both so that users can express our disapproval of their AI features (I noticed when I signed up that the "you used your 100 searches, pay now" page said "let's humanize the web", which from a company that sells AI is pretty audacious), but also because its current features do not adequately comprise an AI opt-out. ("Just don't use it" is, of course, impossible except via the CSS features, because Kagi scatters buttons for its AI features all over the interface; you will eventually hit one by accident.)

Re: Mr. Vlad's post:

"certain technology (that is already opt-in)… Again, Kagi is set up in a way that no AI is pushed to you "

Your AI tech is not opt in, I have already activated the Assistant once by accident because on mobile it's a huge borderless button at the top of every page and to the left of a button one presses regularly. And as noted above AI summaries are mandatory on Kagi News.

"We do not include bullshit in the products"

Your product uses statistical language models to generate text purported to be factual. Your product is bullshit. I resent having to pay for it (not because I'm paying for search; because I'm paying for a plan that explicitly includes "AI" features absent in the cheaper plan), so the least you can do is allow us to exclude from our interfaces features based on machine-generated lies. Making us pay more to turn off the lying is insulting (if it were really true your "AI" is opt-in, then it would also be true that no-AI users cost you the same to support as pro-AI users), and I think you'd get significantly fewer users signing up for an increased cost "no AI" plan than you would for a same-cost "no AI" plan, due to it being so insulting. (I also think if the no-AI plan were even fifty cents cheaper, you'd be shocked to discover how few users will pay even a single penny for AI.) However I personally would pay a no-AI surcharge because I am at this point that desperate, to avoid this non-working pseudo-technology that is being shoved down our throats by every company.

marc-eu

mcc111 However I personally would pay a no-AI surcharge because I am at this point that desperate, to avoid this non-working pseudo-technology that is being shoved down our throats by every company.

There's always Mojeek. Independent search index from the UK. No connection whatsoever with US BT. You have to be able to get used to their strict keyword type of search. You'll hate it at first. But at least it's good old search, with a tiny bit of AI (summarize) which you can turn off and remove from the UI.

Vlad

mcc111

Thank you for approaching this in good faith, although some of the language you use is strong, but not excesively so 🙏

If there were a plan that removed all "AI" capabilities from Kagi, I would sign up for it immediately even if it cost more.

Thanks for acknowledging this, it sets the stage for the right kind of discussion.

Specifically I am currently using Kagi's custom CSS to hide all AI buttons so I do not activate them even by accident.

This futher proves your commitment to it. I'd like to acknowledge that Kagi at least provides a way to do it, even today.

There is a feature request to add another button that hides all AI from UI. It is something we are considering doing.

Which brings me to an interesting point - if there was such an option, would having a $15/mo no-AI plan still be worthwhile having? As you would be able to basically achieve the same thing by toggling this setting on a $10/mo plan?

Additionally, I don't get to use Kagi News because Kagi does not offer any option for viewing news stories except with AI "summaries".

You are probably referring to 'empty' kagi.com/news ? (which is a new thing, and very few people visit)

if you search for news in Kagi eg. https://kagi.com/news?q=ryder+cup none of this is AI generated (at least not by Kagi).

pay now" page said "let's humanize the web", which from a company that sells AI is pretty audacious

I disagree on principle. First of all, AI has been used in search for decades. You wouldn't be able to perform any search today without AI in some form.

Kagi Translate is the world's best translation thanks to AI. People all over the world rely on it to be productive.

I have osteoarthritis in my hands and can input this text using a generative AI model and my voice, instead of typing. Very grateful for that.

I know a blind person who uses generative AI image to voice to 'see the world' and have a much more productive life.

In all these cases, AI is helping us be more human.

I also understand the dangers of AI hype and pushing it on users for the sake of growth, but I'd be surprised if you could accuse Kagi of doing that.

Kagi has been thoughtful about integrating AI from day one. We see it as a technology that has merit when used properly. If someone's distaste for it is 'all or nothing', there's little room for meaningful discussion.

Your AI tech is not opt in, I have already activated the Assistant once by accident because on mobile

Not to be too technical about it, but this just proves my point that it is opt-in (the fact that you accidently activated it does not take away from it). Is it possible to hide it with CSS though? Goes back to my question would a simple remove AI from the UI button be all we need?

Your product uses statistical language models to generate text purported to be factual.

... if the user opts to do so. Not sure if you are talking about Kagi News specifically.

Even then, I believe this is the best way to consume news. Have you tried reading a news website recently? Between ads, tracking, low-quality journalism and opinionated reporting, it's impossible to consume. The way we approach it in Kagi News not only removes all the noise and sticks to the facts, but also offers perspectives on events from diagonally different sources. That's all I care about from journalism, and it would take hours to manually synthesize for just one event.

So instead of being generically prejudiced that "all AI is BS," let's be constructive. Point to an actual example of bullshit created by Kagi News. We worked really hard to ensure it doesn't happen. Not saying it's perfect, but compared to reading about world events on any news website, Kagi News is a 10x better experience.

However I personally would pay a no-AI surcharge because I am at this point that desperate, to avoid this non-working pseudo-technology that is being shoved down our throats by every company.

I hope we are proving that we are not 'every' company by making reasonable choices in product regarding usage of this technology and willingness to listen and directly engage with our customers.

Thank you for taking time to thoughtfully engage.

mcc111

marc-eu Mojeek is interesting in several ways, but they also have a "summarize these results" button on every single search result page. You can turn that off with a setting, but this makes Mojeek in practice pretty similar to Kagi— the company does generative AI, and you can go out of your way to hide it. I guess at least their news feature beats Kagi's on this front (article summaries are lede excerpts rather than being generated text).

marc-eu

mcc111 but this makes Mojeek in practice pretty similar to Kagi

Kagi is going all in with AI. Not Mojeek. The only AI option is summarize, but I give them credit for using a small French/EU service for that.

https://blog.mojeek.com/2024/04/mojeek-search-summary.html

KarolyNegyesi

Vlad

So instead of being generically prejudiced that "all AI is BS," let's be constructive.

AI has been used in search for decades

Sure. Various forms of machine learning produced many interesting and useful results before the fascist billionaires unleashed the scourge named LLMs upon us. This includes search which is technically true but still, it is kind of gaslighting because today AI and LLM are used interchangeably and it is these kind of LLM usages we are so much against.

Gredharm

AI isn't evil.

This is now just semantics. It was created by fascists, some of them self proffered -- it wasn't me who called one of the authors of the Fascist Manifesto one of my patron saints, no, that was Marc Andreessen -- to help with detaching people from reality because that creates an environment where fascism can thrive. So while it might be that the LLM itself is not evil -- after all it's just an algorithm -- it was created with the express purpose of perpetuating evil.

"When we use generative AI, we consent to the appropriation of our intellectual property by data scrapers. We stuff the pockets of oligarchs with even more money. We abet the acceleration of a social media gyre that everyone admits is making life worse. We accept the further degradation of an already degraded educational system. We agree that we would rather deplete our natural resources than make our own art or think our own thoughts. We dig ourselves deeper into crises that have been made worse by technology, from the erosion of electoral democracy to the intensification of climate change. We condone platforms that not only urge children to commit suicide, they instruct them on how to tie the noose. We hand over our autonomy, at the very moment of emerging American fascism."

https://www.nplusonemag.com/issue-51/the-intellectual-situation/large-language-muddle/

marc-eu

Never mind ...

mcc111

I would like to follow up on this.

I have been using Kagi for a few months now. To reiterate, I am using Kagi for one reason, and only one reason: It allows me to disable the "AI" features, and Google does not allow me to disable the "AI" features. I can use the Kagi custom CSS feature to disable the "AI" buttons. As mentioned above I don't believe text-generation "AI" is a real product or a thing that does, or can work; I don't respect people or companies who use or sell "AI"; I am angered when "AI" intrudes on my computing experience, or when I discover something I did triggered an "AI" to gulp some water and belch some CO2. Even though Kagi is an "AI" company, I am giving you money because you are an "AI" company that allows me to hide the AI. No other reason. Your search engine is slower and less featureful than Google, and the ability to hide AI is your only advantage.

Today I was moving my keyboard and accidentally hit some buttons with Kagi up; the Kagi tab disappeared and was replaced with:

I don't know where the letters "bacal" came from, probably keysmash, but that's not the point. I have discovered that if you type the letter "A" while a Kagi tab is up it replaces the entire page with the AI "assistant" and triggers a text generation.

I need to be clear: This is violating. I've gone to great lengths to avoid exposure to AI, I found third-party resources and configured custom CSS to hide all vestiges of AI from the Kagi interface; and you hid a CSS-unblockable feature in the interface to spring AI on me, like a trap, if I accidentally type the letter "A". I now have disabled Kagi keyboard shortcuts completely to prevent the risk of an accidental "A", but now I'm wondering what other ways you will hide AI within the interface and trick me into activating it, which is the exact feeling Google gave me, the thing that I'm paying you money to avoid.

Your AI avoidance features are not enough. The only way you can make AI-vegan customers satisfied is an account level "no generative AI" block, whether that's a different plan type or just a switch. Because you already have features for metering generative AI use, you could easily provide this by simply setting the quota for the account to zero, but you don't, even when customers are begging you for it.

emmatheemu

mcc111

Seems that it's a built in shortcut. While you can prevent this by disabling shortcuts, you do lose ALL shortcuts...
Going back to another thread - I think it would be really nice to have the option to hide all AI features for users that are not interested in them.

mcc111

Further follow ups: After extensive tweaking over months to remove all the "AI" features, y'all just circumvented my CSS and put a "quick answer" slop-generation button on every page.

I'm paying for this search engine. Why do I have to also essentially work a part time job on top of this, identifying and removing each new "AI" feature as you remove them?

It seems reasonable to simply not want AI on my computer. Why do you try so hard to sneak it in?

Thibaultmol

mcc111 tbc, regarding your last message: quick answer has been a feature for years, but the new UX is being changed in an upcoming update. The team is aware of that change not being what the users wanted and is being changed

mcc111

Thibaultmol

Hi, thanks— but to clarify— I don't care at all where the button is. I have a separate problem:

I want for there to be no expressions of generative AI in my interface, and if generative AI is accidentally triggered I want the query to be rejected.
I am currently simulating this by using CSS to hide generative AI panes and buttons.
The change to quick answer breaks my custom CSS because it is a change. It doesn't matter if it is a good change or a bad change— just because it's a change, it breaks the anti-AI CSS. When you change it again (due to "not being what the users wanted") the chances are high you will break the CSS again.
This is not a problem with the "quick answer" button, but rather it's a demonstration of a bigger problem, a problem with AI-opposed users simulating the feature they want (no generative AI) using a feature that doesn't really work (maintaining blocklists of CSS classes corresponding to AI features).

xenofem

I'm in full agreement with mcc111. I want there to be an account-level setting that completely disables any and all genAI functionality in Kagi—no Quick Answer buttons, no "AI assistant" keyboard shortcuts, no genAI content displayed or offered ever unless I go back into my account settings and turn off the "No AI" option. I don't need that to be a separate plan, although I would willingly pay extra for it. If that feature were introduced, I would purchase a Kagi subscription in a heartbeat. Until then, I'm sticking with DuckDuckGo—if I'm stuck with companies pushing genAI either way, then I'll choose the option where I don't have to pay for it.

Gredharm

I really don't understand all the AI hate I see everywhere. AI isn't evil. The companies that supply it, yes they mostly are evil. The governments that fund and promote it and want to use it to spy on and enslave their citizens, yes they are evil. But AI (LLMs specifically) is just fancy auto-complete.

If you hate what AI is doing to the world so much, it's your governments you need to start turning against, not the technology itself.

Remember, every dollar you pay in taxes, is primarily used to enslave you.

larke12

Part of the issue stems from “AI” being an all encompassing term in 2025 for a lot of tools that have existed for years or decades, and modern LLM and generative tools. Most common folk see the high power usage, environmental impact, and copyright-laden issues with generative tools and LLMs and apply that to everything machine learning and AI-adjacent. No amount of simple marketing or explanation can shift that popular perception easily. AI as a label has bad connotations that some people (rightfully in their respective cases) feel strongly about.

There’s nuance to where and how Kagi uses “AI”, how not all machine learning is actually a problem or point of concern, and how people perceive it’s sprinkling all over Kagi’s UI. I’d agree there isn’t a “no-AI” plan Kagi could offer, because the broad label of AI would mean they’d be lying if they offered the same great search just without the assistant, or summaries, etc. Kagi could, however, break down what technologies are used, their impact and cost, so people understand not every tool is a power hungry or content hungry black hole.

I think it’d be hugely beneficial, a boon even, if Kagi offered a default setting to completely disable visible and interactive AI tools. People would also be very willing to pay for a plan where these features are locked away completely. You pay Kagi to use its amazing search, pay for the search infrastructure and not to fund the access to models and non-search AI development. No ability to quick answer, no assistant, no summaries, no translate (though this is publicly available so it’s “free” already, news as well). Kagi was and always will be an AI company, so if people draw the line in the sand at that there’s no way around it for them.

For others, though, just being assured their money isn’t going towards the AI tooling would be enough. It’s easy to connect the higher tier costs for Kagi to additional AI tooling, such as Assistant’s model access. Why pay more for accessing dozens of LLM’s when all a user wants is unlimited search queries? A lot of people just want private and unlimited search, none of the AI bells and whistles, and they’d like to feel that they aren’t directly having to pay for these things they won’t use.

Gredharm

KarolyNegyesi Okay. So use LLMs in a way that benefits you, and sow dissent in the general public until we can overthrow the oligarchs globally.

LLMs are like guns, yes your enemy can use them to kill you... But, what is the best defense against an enemy with a gun? Allies with many guns.

It is true that these people and companies stole vast amounts of data to develop these tools. So the solution isn't to destroy or ignore the tool that now exists and has value. It is to turn against the governments that permitted it and imprison all involved in using it for crimes.

But given a company like Purdue Pharma and the Sacklers went and murdered half a million people in the USA and STILL didn't end up in prison. (Basically paid billions, about HALF the profit they made from mass murder, and in exchange got no criminal liability), I doubt anyone is going to care about big tech companies, without global regime change.

KarolyNegyesi

Gredharm Allies with many guns.

this is the worst kind of 2A rhetoric. The best defense against guns is to create a prosperous environment such that using a gun or many guns is a net negative for the user.

That has nothing to do with the problem at hand though.

And yes I complained about search results before and instead of "yeah we screwed up because it doesn't even search for the words you entered" I got "you are holding it wrong".

OnanForumed

I would like to have the option to pay for unlimited search without any AI interaction features. I don't use these features, and don't appreciate paying for them. I would be more inclined to recommend Kagi to friends, family, and strangers if I could separate the unwanted AI features from the core search product, incentivizing development of one without the other. Please consider offering a lower cost subscription plan with unlimited search that would allow me to pay for AI interaction metered by usage, rather than at any flat rate. If you don't want to do usage-metered pricing, then please consider a low cost tier that only includes search.

In a recent blog entry, you wrote "We’re not planning to force AI onto our users or products. " This appears to be true enough, regarding your UX, but your pricing tells a different story. I understand the economics of LLM services enough to strongly suspect that the costs of these third-party services dwarf the cost of providing plain search, and are largely out of your control. I don't want to subsidize heavy users of AI through an "unused gym membership" pricing structure.

Every review of Kagi's search product that I've seen to date has praised the quality of the search results while grumbling that the $10/month tier is still too expensive for casual users, while the $5/mo tier usage cap is too restrictive. A $5/mo unlimited search-only tier would expand your user base and placate AI-averse users like myself: I know I'm not alone in this.

The search-only home page would look exactly like it does now, but with that "Assistant" logo greyed out. Clicking on it would bring up an upsell modal.

« Previous Page Next Page »