Most search engines are quietly flooding their results with AI-written sludge, but Kagi is betting that a smaller, hand-curated slice of the internet can feel more human than anything scale alone can buy.
A deliberate bet on a smaller, stranger web
In March 2026, Kagi rolled out its Small Web experience to iOS and Android, turning what started as a niche desktop experiment into a way for ordinary users to hop between more than 30,000 non-commercial, human-authored sites from their phones. The move, first detailed by techcrunch.com, is less about chasing raw index size and more about building a counterweight to an internet where AI-written listicles and SEO bait increasingly drown out personal blogs, webcomics and weird one-off projects.
Instead of ranking pages by ad spend or engagement hacks, Kagi leans on a curated index submitted through GitHub and vetted by staff and community members. Coverage on techcrunch.com notes that the Small Web now spans personal blogs, independent video creators, comics and small software projects, all of them required to be fundamentally human-made. That constraint makes the catalog look tiny next to Google, but it also means a user hitting the Small Web next button is far less likely to land on a scraped, auto-generated page pretending to be expertise.
The effect, as reviewers and early adopters quoted by techcrunch.com suggest, is that Kagi search results feel closer to the 2000s-era web, when stumbling into someone’s meticulously maintained niche site felt normal instead of rare. For people exhausted by AI-written product reviews or copycat explainers, that change in texture matters as much as any individual feature.
The AI sludge problem traditional search will not admit
The bet behind Small Web only makes sense if you accept that the mainstream web really does have an authenticity problem. Over the last two years, independent writers and outlets have documented how cheaply generated AI content has flooded search results with pages that look legitimate on first skim but collapse under scrutiny. Those critiques argue that ad-driven search engines have every incentive to keep low-quality, auto-generated pages in circulation as long as they keep users clicking and scrolling.
Reporting on techcrunch.com frames Kagi’s response as a quiet, oppositional move rather than a frontal assault on the giants. By carving out a clearly labeled lane for human-authored sites, the company can down-rank the worst offenders in its main index while giving users a one-click way to opt into something more intentional. That approach does not solve the AI problem for the whole web, but it does give users a real choice instead of vague trust-the-algorithm marketing.
The tension is obvious: Google and other incumbents are racing to bolt generative AI directly into the results page, turning answers into synthesized paragraphs and citing sources only sparingly. Kagi is adding AI as a paid assistant on top of its results, but with Small Web it is also signaling that, for a big class of informational searches, the best answer is still a human who cared enough to build a site about their obsession.
Curated discovery in an era of infinite feeds
Small Web is not just a filter inside the main Kagi search engine. It is also a standalone discovery interface that functions like a modern StumbleUpon: a dedicated site, browser extensions and now mobile apps that let users bounce between random curated pages or browse by category. Techcrunch.com describes category filters for blogs, videos, comics and code, along with a distraction-free reading mode and saving tools so people can build their own mini-directories.
That structure matters because the old discovery tools that once surfaced the independent web have either died or been devalued. Social feeds are dominated by short-form video and engagement-optimized outrage. Algorithmic recommendation systems on major platforms steer audiences toward content that keeps them watching, not necessarily toward the most thoughtful or idiosyncratic work. Against that backdrop, a hand-built list of tens of thousands of oddball sites has outsized cultural weight.
Critics writing about Kagi and the broader small web idea argue that the company still has blind spots: requiring an active feed, for example, automatically excludes many static and archival sites that defined earlier eras of the web. Others question whether a small, privately run catalog can truly represent the breadth of human creativity online. But those complaints mostly concede Small Web’s underlying premise: that someone needs to be doing this curation work at all.
What This Actually Means
The deeper claim behind Small Web is that search quality is not just a ranking problem but a values problem. If your economic model rewards ad impressions at scale, you will keep incentivizing content farms, SEO arbitrage and AI-written filler. If your model depends on people paying directly for better search, you can afford to protect users from that sludge, even when it means sending them to a tiny personal blog instead of a big publisher.
Kagi is still a small player, and nothing in techcrunch.com’s reporting suggests it is about to dent Google’s market share. But it does not have to. The existence of a working, revenue-generating search engine that treats handmade sites as a feature rather than an afterthought is a proof of concept that weakens the usual excuses from larger firms. It shows that you can build a viable business around curation instead of extraction, and that you can use AI tools without turning your results page into a hallucination machine.
What is Kagi and how does its Small Web work?
Kagi is a paid, ad-free search engine based in Palo Alto that positions itself as the inverse of the mainstream, surveillance-driven web. Instead of selling targeted ads, it charges users a subscription fee and promises no tracking, no ad auctions and no behavioral profiling. Public company histories and docs note that Kagi bootstrapped for years before taking modest outside funding, which helps explain its bias toward pleasing subscribers rather than advertisers.
The Small Web sits on top of that model as a separate, curated index. Sites are submitted through an open repository, reviewed against guidelines that prioritize non-commercial, human-authored work and then slotted into topical categories. Kagi’s own documentation explains that the company periodically prunes dead links and spam while adding new submissions, so the catalog evolves over time instead of freezing into a museum of the mid-2020s web.
- The catalog spans tens of thousands of sites, from niche hobby blogs to long-running personal essays and small creative studios.
- Submissions are transparent, allowing anyone to see which domains are being proposed, accepted or rejected.
- The same catalog feeds both the dedicated Small Web browsing experience and a separate signal inside regular Kagi search, helping human-authored sites surface in ordinary queries.
Users retain normal web controls: they can block domains they dislike, bookmark favorites for later and switch back to the broader index when they need something outside the human-authored bubble.
How does Kagi’s business model shape its approach to AI?
Unlike ad-funded giants, Kagi has been blunt in public posts about the trade-offs of adding AI. Company docs and interviews emphasize that AI summaries and assistants are optional, sandboxed layers on top of search results rather than replacements for them. That stance has drawn praise from some privacy advocates but also criticism from writers who worry any AI stack built on web content will end up exploiting the very human authors Kagi claims to protect.
The financial structure matters here. Because Kagi relies on user subscriptions, techcrunch.com notes that the company pitches itself as user-funded and ad-free, which puts pressure on Kagi to show that AI features actually make paying customers’ lives easier instead of just juicing metrics. That is a different set of incentives from platforms that need to keep people scrolling long enough to sell another pre-roll advertisement.
- Subscribers who dislike AI-generated answers can simply ignore them and use Small Web plus traditional results.
- The company’s open communications about investor funding and costs make it easier to see when it is prioritizing growth experiments over its stated mission.
- Because there is no free, ad-supported tier, Kagi does not need to degrade the experience for non-paying users in order to upsell them.
What should users take away from Small Web?
For everyday searchers, the most important lesson is that AI-era search does not have to feel impersonal or extractive. Reporting on techcrunch.com about Kagi’s mobile launch shows a concrete, working example of a search company that treats independent sites as scarce resources worth protecting. The Small Web will never rival a global index for completeness, and it does not try to. Instead it offers a parallel track where curiosity is rewarded with unexpected human voices instead of another stack of AI-regurgitated summaries.
If the experiment succeeds, it could push larger players to treat human-made content as something more than raw training material. If it fails, it will still leave behind a public, open catalog of thousands of independent sites that other projects can build on. Either way, it proves that in an AI-saturated era, there is real demand for search that remembers the web is made of people, not just tokens.
Sources
techcrunch.com, Kagi blog, Kagi Small Web docs, The Small Web is tricky to find, Ars Technica on switching to Kagi