Exactly. Figure out the sort of thing that Google and Yahoo don't or won't do, and do that.
You can certainly leverage their marketing. For the next couple of months, the Flash universe is going to be filled with discussions of search. Google and Yahoo are going to do a bunch of marketing, and a bunch of market research, all of which you don't have to pay for. Just monitor some Flash-user forums, watch the users rave and/or complain about the new tools, and see if a niche market presents itself.
Meanwhile: Launch. Get something out there you can point to.
Notice that what they are suggesting is the key component of your solution (taking links from a page and producing the parts to index) not the indexing itself or the search term parsing &c. I think if you offered that aspect (as a web server or licensed, shared-source software) then you could carve out a good nice whithout having to rebuild the rest of the "search" infrastructure. Basicly, I am suggesting that you forget about
"Our search results are not very good right now. We are in the very early startup phase of our website. The crawler has only been crawling for a while, we only have crawled a handfull of websites, and our search query technology is only in the keyword phase. All of these things require time and money to implement. As things progress we'll improve our infrastructure and ramp-up as quickly as we can."