Hi, I wanted to leave my thoughts after reading the latest at https://kagi.com/changelog#3750
When I discovered Kagi I was elated! Finally, high-quality search that had a reasonable business plan and loaded in under 1s on an iPhone!
Today I am rather sad. Most of the features discussed are not quality goals, but quantity goals. They do not seek to improve tool performance, they seek to ensure titlebars render on more devices and contain AI agents that support search queries I never run.
Right now I have to restart my iPhones safari browser every 6 searches or so, because the API the Kagi extension uses to translate bing.com/search?q=%s into Kagi.com/search?q=%s doesn’t always load.
Search latency has skyrocketed; what used to return to my phone in 1s now takes 4-6s.
I have not noticed any search engine quality dips, but I end up using Kagi less because of the browser incorrectly defaulting to bing more often.
I’d like to ask the Kagi dev team to please please make sure quality remains on your roadmap. It is Kagis differentiator; it’s the reason it’s a valuable tool!
AI is likely going to change the world - but if I know myself it’s as likely to decide the search engine I use as intellisense engines are to decide which IDE I use. Please consider the secondaries of design decisions when planning development; an AI agent has no secondary value agents. A search engine that returns strongly-typed results, and the same results regardless of who typed them in, gives you googles front-end value proposition circa 2004. Without advertising you easily beat googles historic product value.
PS - your feedback mechanism is designed to promote a type of feedback that I have never found useful when building my own tools. I’m a weirdo; I stand behind my customers while they use an alpha tool that fires off exceptions. The quantity metrics have never led me or my team to building something beloved by our customers.
