Within telecom circles, some regard Hossein Eslambolchi as the savior of Ma Bell. In 2002, with AT&T's (T ) long-distance revenues fast disappearing, former Chief Executive Dave Dorman called the company technologist at his Silicon Valley home to ask him to become "chief transformation officer" and push through a radical rewiring.
Three years later, Eslambolchi had replaced AT&Ts (T ) aging, inefficient voice and data networks with a single Internet-based one smart enough to handle everything from basic calls to sophisticated corporate software applications over one set of gear, with less cost and complexity. That was a major reason SBC Communications Inc. paid $16 billion for AT&T in 2005.
Now, having helped design the plumbing that carries hundreds of millions of blog posts, songs, and video clips each day, Eslambolchi is about to launch one of the Web's more intriguing technology forays. His goal? Making sense of all those bits. On Mar. 13, Eslambolchi's 10-person start-up, Divvio Inc., will turn on a service that automatically finds audio, video, and, eventually, text, on your favorite subjects. Then it weaves these clips together to create personalized multimedia channels that are updated each time you sign on. A channel on the New York Yankees might start with spring training highlights, followed by videotaped interviews and blog postings.
That would be quite a change for Netizens, who have grown used to clicking back and forth between bookmarked sites or wading through increasingly cluttered personal portals such as My Yahoo! (YHOO ) or MySpace.com (NWS ). Backed by $3 million-plus raised from angel investors including former executives of Yahoo and eBay, Eslambolchi has been working 16 hours a day since leaving AT&T in late 2005, after turning down an offer to stay with the combined company. "We're going to give you the best of the Web, but only what's relevant to you," says the Iranian-born Eslambolchi, 49.
It's an ambitious goal, backed by technology of mind-numbing complexity. And that might be Divvio's biggest hurdle. Even the most focused startup struggles to solve a single major problem; Eslambolchi is taking on several of them at once, with a technology he calls alip, for "adaptive learning intelligent process."
First, Divvio's software trawls the Web, indexing content not just by subject or title, but by bandwidth, duration, and over a dozen other attributes. Then a "personalization engine" matches each consumer's desires to this data base--sensing, based on past requests, that the user is interested in Madonna the singer, not the mother of Christ. Finally, Eslambolchi says, the Divvio servers will stitch the segments together in milliseconds, regardless of what media format they were coded in.
Sounds impressive. But the site has to work right out of the box if it's going to win fickle Web users. If not, "it could get old very quickly," says Mike McGuire, a Gartner research vice-president. Divvio isn't the only outfit trying to help Web users corral content. For instance, Joost, from the founders of music file-sharing site Kazaa and the Skype (EBAY ) Internet-phone service, is hoping to be a one-stop shop for viewing TV shows over the Net.
Eslambolchi will face a major marketing challenge to break through the Web 2.0 clutter. And early users will need some patience. That's because Divvio will crawl only 750,000 sites at first, a tiny fraction of the entire Web, and because alip needs time to learn. Divvio figures it will only hit the mark 10% of the time for its first customers; he says that will grow to 60% by yearend.
Divvio hopes to make money by sharing ad dollars with content providers, or earning fees with phone carriers, once the service is able to run on mobile devices. After all, if Divvio works, it should generate reams of media-rich transmissions. To Eslambolchi, that would help justify those newfangled networks he had a hand in creating. As he points out: "The world's carriers have invested $10 billion in these [high-speed mobile] networks, but what are they doing with them? E-mail and instant messaging."
By Peter Burrows