Xyo and the foundation for a predictive app discovery engine

By Richard Nieva , written on June 4, 2013

From The News Desk

Here is your meta moment for the day: Predictive apps, like Google Now, Grokr and Tempo, are all the rage. But what if there were a predictive service to help you discover those predictive apps?

Xyo, an app discovery search engine, today launched a product called Apps for Me that works along those lines. Instead of processing a query typed into a search field, Xyo pulls information from everything you’ve "liked" on Facebook and feeds you app recommendations based on those interests. You can also type in the name of any Facebook friend and see the apps recommended for her. It feels a bit like Graph Search for app discovery.

This idea is nothing new. E-commerce runs on recommendations, and has ever since Amazon turned it into an art. Pulling from Facebook interests isn’t anything revolutionary either. The social network does that itself whenever it shows you an ad, just without your prompting it.

I tried it out, and there are still bugs. Because I liked my old undergraduate newspaper, The Santa Clara, Xyo recommended I download the Santa Ynez Valley High School app, a school I’ve never heard of, about two hours away from Los Angeles. Go Pirates? There is also the issue of a user having an under-populated like section, in which case the service recommends a few generic apps based on popularity.

Then there’s the issue of Like pollution, in which your Facebook page is full of things for which you clicked "Like," but don’t really like. Maybe you just clicked the thumbs up to cash in on a deal some brand was advertising. Or maybe you succumbed to the peer pressure of liking your friend’s photography fan page. In which case you’d get recommendations for fashion and photography apps you’re probably not really interested in.

For that, Krzykowski has an answer: “We don’t need to get it right all the time.” If it brings you relevant results about 70 percent of the time, it’s successful by his measures. The point is that you are given a jumpstart in app discovery, instead of having to pull a generic search term like “games” out of thin air.

I’ll give him the benefit of the doubt because his vision for what the technology has the potential to do something interesting. He describes how it can work with wearable tech, like Google Glass, or a smartwatch. Krzykowski says that these devices, mixed with Xyo’s technology, can make app discovery more contextual. For example, you’re wearing augmented reality glasses and you walk past a bookstore. (Let’s be optimistic and say that bookstores still exist.) Then a recommendation for some kind of library or literature app might pop up. This kind of discovery would get tiring if you left that function on all the time, but it would be amusing to turn it on for a few minutes, take a walk, and see what Xyo brings me.

That might seem frivolous, but app discovery in the near future is an interesting proposition. After all, as KPCB partner John Doerr said at PandoMonthly last month, a few apps could be gateways to mobile devices as simulation engines for experiencing things we otherwise might not be able to experience. So a good app discovery engine could be valuable.

Of course, for that app discovery to be worthwhile, the results need to be smarter. And right now, they just aren’t yet. After all, 70% on an exam gets you a C, and that's not a grade most of us would like – on Facebook or anywhere else for that matter.

[Image courtesy: Cristiano Betta]