Hi!

Kagi had a rough couple months on the PR side, and a comment from another Lemmy user arguing that they aren’t using Google’s index set me off… because I had just read a couple weeks ago on their own websites that they primarily use Google’s search index.

Lo and behold, that user was “right”: No mention of Google whatsoever on Kagi’s Search Sources page. If that’s all you had to go off of, you’d be excused for thinking they are only using their internal index to power their web search since that’s what they now strongly imply. The only “reference” to external indexes is this nebulous sentence:

Our search results also include anonymized API calls to all major search result providers worldwide, specialized search engines like Marginalia, and sources of vertical information […]

… Unless one goes to check that pesky Wayback Machine. Here is the same page from March 2024, which I will copy/paste here for posterity:

Search Sources

You can think of Kagi as a “search client,” working like an email client that connects to various indexes and sources, including ours, to find relevant results and package them into a superior, secure, and privacy-respecting search experience, all happening automatically and in a split-second for you.

External

Our data includes anonymized API calls to traditional search indexes like Google, Yandex, Mojeek and Brave, specialized search engines like Marginalia, and sources of vertical information like Wolfram Alpha, Apple, Wikipedia, Open Meteo, Yelp, TripAdvisor and other APIs. Typically every search query on Kagi will call a number of different sources at the same time, all with the purpose of bringing the best possible search results to the user.

For example, when you search for images in Kagi, we use 7 different sources of information (including non-typical sources such as Flickr and Wikipedia Commons), trying to surface the very best image results for your query. The same is also the case for Kagi’s Video/News/Podcasts results.

Internal

But most importantly, we are known for our unique results, coming from our web index (internal name - Teclis) and news index (internal name - TinyGem). Kagi’s indexes provide unique results that help you discover non-commercial websites and “small web” discussions surrounding a particular topic. Kagi’s Teclis and TinyGem indexes are both available as an API.

We do not stop there and we are always trying new things to surface relevant, high-quality results. For example, we recently launched the Kagi Small Web initiative which platforms content from personal blogs and discussions around the web. Discovering high quality content written without the motive of financial gain, gives Kagi’s search results a unique flavor and makes it feel more humane to use.

Of course, running an index is crazy expensive. By their own admission, Teclis is narrowly focused on “non-commercial websites and ‘small web’ discussions”. Mojeek indexes nowhere near enough things to meaningfully compete with Google, and Yandex specializes in the Russosphere. Bing (Google’s only meaningful direct indexing competitor) is not named so I assume they don’t use it. So it’s not a leap to say that Google powers most of English-speaking web searches, just like Bing powers almost all search alternatives such as DDG.

I don’t personally mind that they use Google as an index (it makes the most sense and it’s still the highest-quality one out there IMO, and Kagi can’t compete with Google’s sheer capital on the indexing front). But I do mind a lot that they aren’t being transparent about it anymore. This is very shady and misleading, which is a shame because Kagi otherwise provides a valuable and higher quality service than Google’s free search does.

  • Dark ArcAEnglish
    91·
    7 months ago

    Ironically going to use Kagi to summarize the blog post:

    Kagi is trying to expand into too many different products and services beyond just their search engine, which is stretching their resources too thin.

    Kagi spent a significant portion of their funding (1/3) to set up a t-shirt printing business to give away free t-shirts to their first 20,000 users, which seems like a questionable financial decision.

    Kagi was not paying sales tax for two years and had to retroactively pay up, indicating potential financial mismanagement.

    Kagi is heavily focused on developing AI tools and features, to the point where the author believes it is becoming the main focus over improving the core search functionality.

    The author is very critical of Kagi’s founder Vlad’s dismissive attitude towards privacy concerns and his belief that email addresses are not personally identifiable information.

    Vlad has a “my way or the highway” management style and is unwilling to consider feedback or criticism about Kagi’s direction.

    Kagi’s dedication to privacy is questionable, as the founder does not seem to take many privacy concerns seriously.

    The author believes Kagi’s AI-powered features like “FastGPT” and the “Universal Summarizer” are inaccurate and unreliable.

    The author is skeptical about Kagi’s long-term sustainability and viability as a business.

    Overall, the author has lost faith in Kagi due to the company’s questionable financial decisions, overreliance on AI, and the founder’s dismissive attitude towards user concerns.

    I did actually read the post (and I think this is actually my second time seeing this). I’m majorly unconvinced by the author … and yes, the criticized AI summarizer is that good. I regularly use it after reading something to share the details with friends (or get a rough idea of what’s being discussed and decide if I want to read something non-trivially long).

    It also works on YouTube videos (presumably using the transcript) which can be a HUGE time saver.

    Ultimately the search is good; it’s better than what Google offers me, and I’ve found their AI tools fairly useful (despite having distrust for GPT-style chat bots/BS generating AI, I think summerization of some specified source is something they might actually do well – the major concern of piecing together random pieces of random sources of varying integrity is largely mitigated).

    Whether the company will stay private / whether it lives on beyond Vlad is the biggest concern I have with using it. However, “what’s the other (practical) option to invest in?” I find myself in a similar position with Steam and Proton (at least the latter open sources much of their work). For now anyways, the weather is fair, so I’ll stay on board.