Having jumped into the market for Web "discovery" technology last December, Aggregate Knowledge is getting closer to general availability of three new services. Founded by the same guys behind social networking site Tribe.net -- the assets of which were acquired by Cisco in March -- Aggregate Knowledge promises to help companies get the most appropriate products and content in front of Web site visitors.Since it was formed two years ago by Paul Martino (CEO) and Chris Law (VP of products), Aggregate Knowledge has banked $25 million through two rounds of financing and is in various levels of engagement with 30 customers, including a half dozen paying customers. Overstock.com and WashingtonPost.com are reference accounts.

Kleiner, Perkins, Caufield & Byers announced in December that it was making a $5 million investment in Aggregate Knowledge, and since then the company has been filling out its management team. In the spring, it named a VP of engineering (Joe Ghazal), chief scientist (Tim Vogel), CFO (Steve Pantelick), and CMO (Dave Peterson).

The company's technology, offered as a hosted service, is an algorithm-powered recommendation engine that collects Web traffic data, analyzes patterns of user behavior, then pushes content or products to Web site visitors based on what like-minded people have done. CMO Peterson calls it "serendipity," or the accidental discovery of something valued but not necessarily sought. Compare that to search engines, he says, which too often fail to find what users are looking for.

In the fourth quarter, Aggregate Knowledge will formally launch three services that are in limited availability today. They are a "discovery network" that extends the company's functionality across multiple Web sites, the capability to use its discovery engine in e-mail marketing campaigns, and a service to support links between a Web site and affiliates.

Aggregate Knowledge is going after media companies and retailers as the low-hanging fruit among potential customers. Other verticals could come later.

Does it work? It must in order for Aggregate Knowledge to make money. The business model, Peterson says, is "pay for performance." For retail accounts, Aggregate Knowledge takes a cut of increased sales. For media companies, fees are tied to associated increases in site traffic.