Search for everyone

A paid, ad‑free search engine, Kagi, has introduced a limited free tier (50 searches without an account, then 100 more with signup), prompting debate over whether its cleaner results and integrated AI assistant justify a subscription in an era where many people now rely on general‑purpose chatbots or are satisfied with DuckDuckGo and Google. Fans highlight Kagi’s high signal‑to‑noise ratio, customization features, and resistance to “AI slop,” but recurring concerns include pricing, limited maps/local search, Cloudflare tracking, and the lack of bundled API usage for agents. A significant ethical objection is Kagi’s paid use of Yandex and other opaque third‑party indexes, which some view as indirectly funding Russia’s war effort and grounds for boycotting the service.

Perceived Benefits of Kagi

  • Many subscribers say Kagi feels like “Google when it was good”: higher signal‑to‑noise, less SEO/AI slop, better surfacing of small blogs and obscure prior art.
  • Ad‑free results are framed as a real quality‑of‑life improvement and important for aligning incentives with users rather than advertisers.
  • Users praise customization: lenses, downranking or blocking specific domains (e.g., Pinterest, Reddit, AI garbage, listicles), and “bangs” for direct site search.
  • Smallweb and curated features are appreciated for replacing doomscrolling with exploring personal sites.
  • The integrated AI assistant is seen by many as well‑tuned to search, with easy on‑demand activation (e.g., via “?”) instead of forced summaries.

Critiques and Limitations

  • Local search and maps are repeatedly cited as weak: missing/moved businesses, poor location detection, lack of filters, limited transit info, weak mobile integration.
  • Some find search results slower or occasionally outdated vs Google, especially for niche or very fresh content.
  • Android/iOS integration is uneven: Safari/extension behavior breaks across OS updates; default‑search setup on mobile is cumbersome; some report a buggy Android app.
  • Result depth is limited (≈200 results); power users want more pages and deeper, less‑curated “old web” exploration.

Pricing, Plans, and API Debates

  • Opinions vary on value: some happily pay $10–$14/month (or family/duo discounts); others want a cheaper “search‑only, no AI” plan around $3–7 or per‑search billing.
  • Heavy AI‑assistant users complain about burning through included AI credits; light users feel they subsidize AI they don’t want.
  • Many argue that API/agentic search should be bundled with subscriptions; Kagi staff/defenders counter that automated use can 100× costs, so PAYG API pricing is necessary.

Ethical/Political Concerns (Yandex)

  • A recurring, heated theme is Kagi’s use of Yandex as one of many data sources.
  • Critics argue any paid integration with Russian companies indirectly funds an aggressive war and is unacceptable; some cancelled subscriptions over this.
  • Defenders say Yandex is a small cost share, that boycotting all Russian firms is overbroad or symbolic, and that pushing politicians matters more than targeting Kagi.

“Search for Everyone” Free Tier & Privacy

  • New no‑account free usage (≈50 searches, plus 100 after signup) is welcomed as lowering friction for trying Kagi and for fallback use when logged out.
  • Some see it as marketing spend; others criticize the limit as still making it effectively paid‑only long‑term.
  • There are concerns about Cloudflare Turnstile, perceived browser fingerprinting, and the tension between privacy branding and this anti‑bot setup.
  • Admins request DNS‑enforced SafeSearch endpoints; absence of this is seen as a blocker for schools and locked‑down networks.

Kagi vs Other Search and AI Tools

  • Comparisons:
    • DDG: often “good enough” and free; some see Kagi clearly better, others see little difference.
    • Google: still best for local and obscure coverage, but widely criticized for ads, “enshittification,” and unreliable AI overviews.
    • Alternatives mentioned: SearXNG, Brave Search, Uruky, Ecosia, Qwant, Marginalia, Exa.
  • Several users say they now do most information retrieval via LLMs (ChatGPT, Claude, Gemini), using search only as a backup—or using Kagi as the search back‑end for their own agents.