You need to find an actual niche that solves a real problem people have and can understand and orient everything you do to tackling that. Then expand from there.
I think there are two distinct things here:
1) Searching the whole internet
2) Returning results that aren't necessarily from the Internet, but instead are convenience features of the engine
I understand that you're not trying to replicate things like "What's the weather today", but when I want results about <very specific classic car X>, how can you return meaningful results without searching the whole Internet?
Put another way, if you don't search the whole Internet, the results are going to be limited to only the curated list of sources you do search. This can be useful in its own way - i.e. if you are positioning this as "search this list of curated sources", but also means the site will only be as useful as the curation you provide.
For example, I dabble with Software Defined Radio. If I search your site for "rtlsdr", a very popular package, I get three results. Those results are somewhat interesting, but I know there's a whole world of content out there related to rtlsdr that I'm not seeing here.
So adding a bit to what the parent commenter was saying - if I'm using your site to look for my particular niche, and I only see three results when I know there are many more, I'm not likely to continue using your site to search for rtlsdr.
It then leads me to wonder what I can search for, or if there's much utility to searching at all.
Please take these comments in the spirit they are intended - I think a search engine that helps find things on the "old" web, or just helps me cut through all of the SEO optimized crap is a great idea. It's something I want to use. But I can also understand why someone might try a search and move on.
Just an idea, but maybe providing a way for independent creators to submit their site for indexing (or for an interested user like me to submit a site) would help increase your reach.