Independent search, built by Nic Weyand
Find the page. Read the source.
Never Betray the User
Argand is an independent search engine for the open web. It crawls and ranks its own index, sends result clicks to the sites that published the work, and does not replace those pages with generated answers.
I am building it in Rust, on hardware Argand controls. The beta is live, the index is still growing, and weak searches are treated as bugs to fix.
Early beta. Coverage and ranking are still improving.
The operating principle
The product should work for the person using it.
That principle is enforced with four concrete rules. They shape the index, the ranking system, the business model, and the data Argand keeps.
Organic ranking has no sponsor or bid input. An advertiser cannot pay to move a page higher in the results.
Tap, swipe, or use the arrow keys.
How it works
From query to ranked sources.
Argand combines exact text matching, related-language retrieval, and semantic retrieval. It merges the candidates, reranks the strongest pages, and uses Navigate only when it can identify a reliable passage on the source page.
Argand normalizes spelling, punctuation, and obvious shorthand while preserving the terms that carry the meaning of the query.
Tap, swipe, or use the arrow keys.
Source-first search
The result is a path to the source.
Argand ranks existing pages and opens the one you choose. It does not put a generated summary between the reader and the publisher.
Argand's live query path runs on ordinary processors. It does not call a large language model or require a graphics card to answer a search.
Tap, swipe, or use the arrow keys.
Different by design
Four differences that affect the result.
- Organic rank is not for sale
Sponsored placement is separate and cannot change the organic order.
- Sources remain the destination
Argand opens the publisher's page instead of generating a replacement answer.
- No behavioural search profile
Anonymous queries are not joined into a history used to personalize later results.
- Independent crawl and index
Argand gathers and ranks its own corpus instead of reselling another engine's results.
Details and evidence
Check the claims.
These pages explain the measurements, policies, current limitations, and release history behind the homepage.