Trapit Uses Siri AI For Content Discovery On iPad
How many Flipboard, Pulse, and Zite competitors can one platform handle? How many “personalization” promises can we stand?
The question is rhetorical. We don’t know yet, because the flow of new players just won’t stop. This last week a new content discovery engine launched its iPad version, claiming to use the same artificial intelligence and learning engine as Apple’s Siri. Trapit has been on the Web for about half a year, but its free iPad version is an exercise in touch interface design as well as some apparent personalization smarts.
Probably closer to CNN-owned Zite than to Flipboard or Pulse, Trapit is not relying on RSS feeds or social graphs to surface content that it thinks interests you. This app sources over 100,000 information sources on the Web, but requires some basic declaration of interests by you. You enter a search term and it “traps” relevant information. It also prompts you persistently to offer more input and a thumbs up or down on items, to teach the system what you do and don’t want. Over time your article selection and like and dislike declarations are supposed to refine the feed.
The app has a superb interface in its favor. In some ways more compelling than Flipboard’s media analogue of the magazine, Trapit goes with walls of interactive tiles. You tap to enlarge any story and then tap again to fade in a full-screen view. A large thumbnail is accompanied by the lead graph of the story, but as you determine whether this is a story worth reading in full the lower part of the screen is already pulling in the full Web page. In this way, Trapit avoids some of the controversy that both Flipboard and Zite encountered when they scraped content from source sites, robbing them of their ads. This interface pulls in the full, original page. There is a good Reading List function for clipping pieces and for quick sharing.
It is slick, to be sure -- and it allows for trapping well virtually any search term combinations, although others do this too. The real issue for personalization apps is determining their overall accuracy and quality. After all, they are scraping a massive pool of available data. There is no way of knowing what they miss. I run Flipboard, Pulse, and Zite regularly, and after months of regular use I still can’t determine which really provides the best feed. In other words, no one of them has floated to the top.
I have to say that while I Iike the depth of Flipboard, I prefer the Zite interface now. But on a qualitative basis, there is a fundamental challenge for AI-driven personalization. Aside from interface, it is hard to communicate superior value to the consumer when the experience is designed to feel seamless and subtle and there are so many ways of creating a “personalized” window into information.