From the same artificial intelligence platform as Siri comes its cousin Trapit, a personalized web search tool powered by one of today’s most popular technologies. It launches its public beta today, letting in the onslaught of users anxious to see the next iteration of AI’s personal assistant toolbox. Trapit has been in private beta for several months now, testing a system that learns your preferences around web content based on your likes, dislikes and search activity. It’s entering an extremely crowded market with a known AI system to differentiate itself, and taking advantage of budding consumer interest around similar AI products.
If you’ve been privileged enough to gain early access to Trapit, you’ll immediately recognize some of the changes pushed through with the open beta. Trapit’s user interface functions much more like a search engine, encouraging you to hone in on your interests based on keyword, phrase or URL. From there you get a series of recommended articles from a variety of manually-approved, clean sources. Click through to the original source, thumbs up or down it, and share the article with friends on Facebook and Twitter. The URL search tool in particular is a key aspect of Trapit’s technology, as it extracts the primary content of an article for its unique characteristics, instead of the more literal approach Google takes. I’m hoping it will work quite well as a supplemental discovery tool for unearthing additional sources for my Google Reader.
But is Trapit’s core technology enough to help it truly stand out from its competitors? Trapit has about 100k clean sources to search in real-time, readily updating your results. There’s no pre-defined topics, so you really end up creating your own channels within your Trapit account, wholly avoiding categorization of the web. Running on two datacenters split between Atlanta and Torronto, Trapit has a fully redundant webscale platform based on Cassandra and Lucene. Trapit’s not using big relational databases, but a process more akin to social networks, built for speed. This is how Trapit plans to stay lean enough to scale its real-time offerings, but what about its core business?
“Lots of companies are throwing around semantic search, but we look at this whole new field of discovery across two axes,” Trapit co-founders Gary Griffiths and Henry “Hank” Nothaft, Jr. explain. “First there’s precision–what you expect to get when you do a search. You’re looking for a fact. Secondly, there’s serendipity–discovery. We’re blending these two. You see an article that’s interesting and you want to find more.”
This balance could provide a nice environment for publishers and targeted advertising, which are two areas of potential interest for Trapit’s monetization. During the beta period Trapit saw a high percentage of stories people are “trapping” and being shared because of their content originality. In this way Trapit is looking to expose more content from its database of sources. Griffiths and Nothaft also indicated an upcoming API to lure in developers that could create additional products and services around Trapit’s engine.
The API was one of the more promising points brought up during my chat with the Trapit co-founders, because the early beta release is lacking extended points of integration and a mobile strategy. Adding the ability to port your Trapit activity feed to another reader or “newstand” app could ultimately increase user activity by extending its distribution channels. Likewise, incorporating more “actionable” functions around its content would better contextualize Trapit’s search, which is still rather isolated as a database of sources. Griffiths and Nothaft are already in talks with a popular iPad news reader in an effort to expand its mobile offerings, and plan to flesh out some of these other points of interaction around its content as well.
Two of Trapit’s biggest strengths are its base technology and its passion for data, seeking the perfect balance between search results and recommendations. It’s a tough line to walk, but one that’s becoming increasingly important in the area of search, largely due to its social and real-time contributions. Trapit has a long list of upcoming features that will only add more functionality to a system that’s always teaching itself, and in this new era of Siri, we’re certain to see a growing interest for AI platforms from consumers and investors alike.