Venky Harinarayan on Kosmix and Vertical Search (Part 2)
Check other articles in the series...SM: Where did you get the idea for your current venture? What is your domain experience in the segment?
VH: We started with the thesis that the Web is like the library of Alexandria – the repository of all human knowledge. The only window people have into this repository today is search. Our vision was to have programs accessing this Web intelligence and mining it to create new applications. Since search returns an unstructured set of 10 results, programs cannot be built on top of it.
Fundamental to building such a Web intelligence platform is the need to “categorize the Web” — structure the Web into categories. These categories become the handles on which programs operate that humans can use in their day-to-day interactions. Much of this came from the search and domain experience we had cultivated during our tenure at Junglee and Amazon.
SM: What was the market landscape like when you founded the company? Competition? Competitive Positioning?
VH: We build “home pages” for topics as varied as “rooster comb injections” to “Acura NSX”. We do so algorithmically, by using Web intelligence to lay out and “program” the page. While no one is doing exactly what we’re doing today, our business model is most similar to that of a content site. From an operational point of view, we find ourselves focusing most on organizations like WIkipedia, Mahalo, WebMD, Edmunds, etc.
SM: Describe the value proposition, including differentiation versus the rest of the market.
VH: There is an explosion of content especially that created by users today. Yet, many of them sit in silos on the Web. For instance, a topic like “diabetes,” has information in mainstream media sites, blogs, social networks (Facebook and MySpace)), videos, UGC content sites, message boards, news feeds and the list goes on. There is no single place on the Web where you can go to see what’s happening on the Web for this topic. We want to be that place – the unofficial home page for every topic on the Web. Think of us as a starting point from which users can explore the Web.
Other players do this manually, so scalability in terms of topics covered is miniscule. Mahalo, for example, returns tens of thousands of topics. We do tens of millions.
SM: Sounds like what About.com had set out to do manually!
(to be continued)
This segment is part 2 in a 4 part series Segment 3 →
Jump to part: 1, 2, 3, 4
