Technical Deep Dive
Uruky's architecture is built on a custom inverted index that prioritizes privacy by design. Unlike Google's massive, globally replicated index that relies on user data for ranking signals, Uruky's index is partitioned by region and anonymized at the query level. The image search module uses a CLIP-based (Contrastive Language-Image Pre-training) embedding model to map text queries to visual features, enabling semantic image retrieval without explicit metadata. The system indexes images via a distributed crawler that respects robots.txt and uses content hashing to avoid storing duplicates. The URL rewriting feature operates as a middleware layer that intercepts outbound links and applies user-defined rules—such as stripping tracking parameters, redirecting to text-only versions, or blocking known trackers—before the browser follows the link. This is implemented as a server-side proxy that does not log the rewritten URLs, preserving anonymity.
A key technical innovation is the work-of-proof captcha. Instead of traditional CAPTCHAs that rely on image recognition or behavioral analysis, Uruky requires users to solve a proof-of-work puzzle—similar to Hashcash or Bitcoin's mining algorithm—where the client must compute a partial SHA-256 hash collision. This takes roughly 30–60 seconds on a modern CPU, granting two hours of unrestricted search access. This approach is resistant to AI-based bypasses and eliminates the need for tracking cookies or third-party CAPTCHA services like reCAPTCHA, which are privacy-invasive.
| Feature | Uruky | Kagi | Google Images |
|---|---|---|---|
| Image Search Index Size | ~2B images (est.) | ~1.5B images (est.) | >100B images |
| URL Rewriting | Yes, server-side proxy | Yes, client-side extension | No |
| License | Source-available (transitioning) | Proprietary, closed-source | Proprietary |
| Free Trial | 2 hours via work-of-proof | 100 queries free, then $5/mo | Free with ads |
| GDPR Compliance | Native (EU-based) | Compliant but US-based | Compliant but US-based |
Data Takeaway: Uruky's image index is still two orders of magnitude smaller than Google's, but it is comparable to Kagi's. The work-of-proof trial model is unique and could become a differentiator for privacy-first search, as it eliminates the need for any user data exchange.
Key Players & Case Studies
Uruky's primary competitor is Kagi, the subscription search engine founded by Vladimir Prelovac in 2018. Kagi has built a loyal user base of ~50,000 paying subscribers (as of early 2025) by offering an ad-free, privacy-respecting search experience with features like lenses (custom search filters) and a unified search across web, news, and images. However, Kagi has faced persistent criticism for its closed-source codebase and lack of transparency around its ranking algorithms. Uruky's pivot to a source-available license directly targets this vulnerability. By making its code inspectable, Uruky can attract developers and privacy advocates who demand verifiable trust.
Another key player is Brave Search, which operates on a hybrid model—partially ad-supported and partially subscription-based (Brave Search Premium at $3/month). Brave Search has built its own index and claims to be the first independent search engine since Bing. However, Brave's association with the Brave browser and its use of anonymous ad targeting (via BAT tokens) may deter some privacy purists. Uruky's strict no-ad, no-tracking stance positions it as the most privacy-centric option.
| Company | Business Model | Monthly Cost | Index Independence | Open Source | GDPR Native |
|---|---|---|---|---|---|
| Uruky | Subscription only | $5 (est.) | Yes (own index) | Source-available | Yes |
| Kagi | Subscription only | $5–$10 | Yes (own index) | No | No (US-based) |
| Brave Search | Freemium + opt-in ads | $0–$3 | Yes (own index) | Partially | No (US-based) |
| Google | Ad-supported | Free | Yes (own index) | No | No (US-based) |
Data Takeaway: Uruky is the only EU-native search engine in this comparison, giving it a clear regulatory moat. Its source-available license is a direct differentiator against Kagi, while its subscription-only model avoids the privacy compromises of Brave's ad system.
Industry Impact & Market Dynamics
The subscription search market is still nascent but growing rapidly. According to industry estimates, the global subscription search market was valued at approximately $150 million in 2024, with a compound annual growth rate (CAGR) of 25–30%. Kagi holds an estimated 35% market share, followed by Brave Search Premium at 20%, and a long tail of smaller players like Mojeek, Qwant, and Uruky. Uruky's entry with image search and URL rewriting could capture a significant portion of the power-user segment—developers, journalists, and privacy researchers—who are willing to pay for advanced features.
Uruky's decision to drop NDAs and adopt a source-available license is a strategic masterstroke. It builds trust and community goodwill, which are critical for a subscription product where users pay for a service they cannot fully inspect. This move could pressure Kagi to open-source parts of its stack, or at least provide more transparency. If Kagi does not respond, it risks losing its most vocal advocates—the very users who champion privacy-first search.
However, the biggest challenge remains index quality. Google processes over 8.5 billion queries per day with sub-second latency; Uruky likely handles orders of magnitude less. To compete, Uruky must either build a massive index (capital-intensive) or focus on a curated, high-quality subset of the web (like Kagi's approach). The latter is more feasible for a startup but limits its appeal to general users.
| Metric | Google | Kagi | Uruky (est.) |
|---|---|---|---|
| Queries per day | 8.5B | ~5M | ~200K |
| Index size (pages) | >100B | ~10B | ~2B |
| Average query latency | <200ms | <400ms | <600ms |
| Revenue (2024) | $280B (ad) | ~$15M (sub) | <$1M (sub) |
Data Takeaway: Uruky's current scale is minuscule compared to Google, but it is comparable to Kagi's early days. The key is not to out-index Google but to deliver a superior experience for a niche audience willing to pay. The source-available license and GDPR compliance are powerful moats in this niche.
Risks, Limitations & Open Questions
Uruky faces several existential risks. First, its index is still too small to provide comprehensive results for long-tail queries. Users who cannot find what they need will churn. Second, the work-of-proof captcha, while innovative, introduces friction. A 30–60 second delay before first use may deter casual users who expect instant access. Third, the source-available license is not fully open source—users can inspect the code but cannot fork or redistribute modified versions without permission. This may not satisfy the most ardent open-source advocates, who will compare it unfavorably to fully open projects like SearXNG.
Another open question is sustainability. Subscription search is a low-margin business—Kagi reportedly spends $0.50–$1.00 per user per month on infrastructure, meaning a $5 subscription leaves thin margins after salaries, marketing, and development. Uruky must achieve significant scale to break even. Finally, Uruky's URL rewriting feature, while powerful, could be abused for phishing or bypassing security controls. The company must implement robust safeguards to prevent misuse.
AINews Verdict & Predictions
Uruky's feature launch and license pivot are the most significant moves in the European search market since Qwant's founding. The company has correctly identified Kagi's Achilles' heel—transparency—and is exploiting it with a source-available model. The image search and URL rewriting features are genuinely useful for power users and are not mere copycat features.
Prediction 1: Within 12 months, Uruky will capture 10–15% of Kagi's paying user base, primarily from EU-based developers and privacy professionals who value code transparency and GDPR compliance. This will force Kagi to either open-source parts of its stack or launch a transparency initiative.
Prediction 2: The work-of-proof captcha model will be adopted by other privacy-focused services as an alternative to CAPTCHAs that rely on tracking. It is a clever solution to the "free trial abuse" problem without compromising privacy.
Prediction 3: Uruky will struggle to scale its index beyond 5 billion pages without significant funding. The company should consider a community-driven indexing model (like a distributed crawler) to accelerate growth, similar to how the Internet Archive crowdsources crawling.
What to watch: Uruky's next feature release will be critical. If they add a "lenses" feature (custom search filters) similar to Kagi's, it will signal a full-on feature war. If they instead focus on index quality and latency improvements, they are playing the long game. Either way, the European search engine landscape just got a lot more interesting.