As others have said, I don't think image generation is a good fit as a general concept.
The comparison with summarizing text isn't fully fare in my opinion. For me, search is about finding what you're looking for, not creating it. By summarizing a page I can find the information I'm after quicker than reading the whole thing, for example.
The closest thing I can think of when it comes to image generation is something like "What does X look like?" to find out whatever you're looking for looks like. For music/sound it could be something like "What does X sound like?" and for those kinds of use cases I suspect generative AIs are too immature and would risk hallucinating too much. If one can have references like the assistant it should be fine though. When image and sound generation is more reliable it might be a better idea though.
But a general content generator doesn't fit in my opinion. Like, why not build an image editor into Kagi? Because it solves a different problem, just like an image generator.
I don't hate the idea in general and I wouldn't boycott Kagi if it was implemented, but I'd rather see focus and effort spent on things related to searching and finding information rather than competing with GenAI services directly. If it's low effort to implement and it doesn't affect what I pay I don't mind it.
