This is the kind of feature that would potentially make me switch, but I feel like at this point AI has disrupted web search for me in such a fundamental way that I rarely use a search engine nowadays.

Just be careful with nuance. Someone close to me asks Gemini absolutely everything because they like how it gives you a straight answer, but real life doesn't always work that way. Context is often missing (besides just being wrong at times).

Search engines shouldn't be replaced by AI but they unfortunately are and it result sucks to see.

I'm still getting there, but I've seen this. I spent hours a couple days ago trying to get pi, llamacpp, and qwen3.6 working well. I had them working before, but not well. I Googled, and I followed it's instructions and following the links it found. But everything was a partial answer, and nothing was clear.

I finally decided to ask Claude to do it. It discovered things I didn't even know to look for, immediately making it run twice as fast, and then also helped me do the changes I wanted, like per-model settings.

I knew it was settings.json, but I had it in the wrong spot (thanks, Google) and what I didn't realize is that the 2 settings.json files are different. 1 is global-only, and 1 is per-model-only. If you put the wrong settings in, it'll just ignore them silently. Which left me wondering why it didn't work.

Then it tested my settings. Then I had to use the official webpages to set the proper settings and it tested those, too. It made suggestions about removing or changing some settings that also made things work better and faster.

In less than a hour, it had vastly outstripped everything I'd tried, and it got it right and verified it.

And with how wrong I've seen Google's AI be (and how wrong others have said it is) I can't trust it to even bring me the basics about things. It's worse than useless, it's distracting.

Kagi also offers Kagi Assistant, which seems to harness models to apply a search-first behavior, and tends to produce higher-quality results for information that would not normally be within a given model's training data. While the cheaper Kagi plans don't offer the full slate of frontier models, I've found that Kimi models in Kagi Assistant are generally excellent for discovering information, even if I may then go to Claude or Gemini frontier models to parse what I've found.

Kagi Assistant also seems to be a bit lighter on guardrails. I was trying to understand how Libby ebooks are stored and encrypted a few weeks back, and no matter what I did, I could not get any Anthropic products to help me - they insisted I was an evil pirate. Kagi Assistant with either Kimi or Grok (sigh) were happy to help. (If I was doing something nefarious, not sure how far they would have gone, though.)

There are quite a few Luddites like myself that don't trust AI summaries and would rather just have the relevant links, so for us this would be useful. Though TBH I'd rather just stick with the paywalled articles being highlighted as such so I can choose on-the-fly. Perhaps the block option is something I'll turn on when almost all the results are paywalled and I want to see if there is a better option before going to one of them.