You say Per-say, I say Persai - let’s call the whole thing off

It hard to separate news aggregator start-up Persai from the personality of one of its notoriously pugilistic co-founders, Ted Dziuba. Before Uncov, however, Ted got a degree in computational math and worked at Google, and co-contributors Kyle and Matt certainly seem to bring deep knowledge to the table: both the Persai blog and Dziuba’s former blog (Epsilon-Delta) at least have enough mathematical jargon to wrap an air of credibility around the enterprise.

So - what is Persai? At a high-level, as I understand it (based on the “evidence” cataloged below), its a news reader/aggregator, similar to in basic appearance to Google Reader. Where Persai differs, apparently, is in the fact that your reader’s content is determined by your interests - not just feeds you’ve subscribed to. Persai, based on your added interests and its own relevance engine, serves up content for you in a Reader-like form - acting as a “recommendation engine.” You can then further refine Persai’s interpretation of your interests by rejecting its selections, creating what I imagine to be a Pandora-like experience. Ultimately, it looks like Persai is closer conceptually to an uber-Memeorandum, Megite, or Tailrank - offering a more customizeable, granular experience integrated into a single personalized stream, compared to other sites canned verticals.

There are three pieces of Persai out there right now: a screenshot at Valleywag (displayed above), and two microsites (eyeonfacebook and appleinsight) apparently built to demonstrate Persai’s ability to generate a topical content stream. There’s also the Persai blog, which is long on technical descriptions and short on use-cases.

So - given that I don’t have access to a beta or whathaveyou, I can’t comment on how well it works, or how accurate my understanding of its functionality is. Assuming that I do have generally the right idea, however, I can comment on the business elements of Persai.

In a nutshell: I think Persai’s user-customizeable recommendation engine is bound to secure a dedicated niche user base, and less likely to ever go “mainstream.” Interest-based recommendations would certainly make an interesting addition to Google reader, especially when combined with your search and Reader clickstream history. As a standalone, however, I’d expect it to float in the same orbit at Megite/Tailrank/etc. - perennially there, but never crossing the threshold that separates those businesses that exist from those that win.

This entry was posted on Monday, December 17th, 2007 and is filed under Site Profiles. You can follow any responses to this entry through the RSS 2.0 feed.