[intrasight]

Doesn't Kagi just use search APIs to get it results? Why would they be better if they don't build their own index?

[thefurdrake]

Kagi lets me block domains from ever showing up again on my searches ever. That alone is worth 5 bucks a month to me.

[dabluecaboose]

Wow, I gotta be honest it might be worth $5/mo to never accidentally get sent to Pinterest ever again

[thefurdrake]

Or fuckin' geeksforgeeks.

I am totally not affiliated with kagi but since I started using it, I haven't gotten pissed at my search engine for sucking ass.

OH! Another feature that actually works on Kagi: Date filtering! Never again will you set that date filter to the last two weeks and receive results that were posted in 2005.

[GrinningFool]

https://kagi.com/faq#Where-are-your-results-coming-from

[alx__]

Where have you heard this?

They build their own index: https://help.kagi.com/kagi/why-kagi/kagi-vs-competition.html

[J_Shelby_J]

They build their own index but don’t let you access it with keyword based queries. Instead the do ‘magic’ and shape the responses from multiple data sources.

I’d love it if instead of ‘magic’ we just had a search engine that let you be the magician with a better query language and ux filters.

[VHRanger]

They're probably doing something like mapping the embedding of the query to the nearest neighbor in the search given how the front-end of their search infra works

[CalebJohn]

Kagi uses multiple search APIs (including their own) and implements their own ranking and mixing of results. Basically you get the same results with spam removed, with some truly unique results from the in-house engine.